ABSTRACT

"CROSS-FIELD" VISUAL MASKING BY TACHISTOSCOPIC PRESENTATION OF
TARGET AND NOISE PATTERNS TO OPPOSITE CEREBRAL HEMISPHERES

BY
David P. Goff

Methods for studying hemispheric asymmetry in the processing of
sensory information by normal human subjects involve unilateral presenta-
tion of stimuli which then project to the opposite hemisphere. In the
employment of these techniques, detection of laterality differences is
often difficult. A reason for this is that several milliseconds after
initial projection to one hemisphere, information becomes available for
processing by the other hemisphere due to transfer across the commissures.

The goals of this study were to develop a new technique for increas-
ing observed laterality differences in the visual modality and test the
underlying explanatory constructs. The procedure employed tachistoscopic
presentation of target information to one hemisphere and a pattern "mask"
to the other. Although target and mask information then transfer to
apposite hemispheres, it was predicted that the mask would produce less
interference in one hemisphere than in the other due to attenuation through
the corpus callosum. Consequently, a greater laterality difference would
be observed under masking than with conventional procedures.

Target stimuli were single alphabetic letters, and recognition
accuracy was the dependent variable. When the pattern was projected
initially to the left (language) hemisphere, masking should have been

greater than when mask was projected to the right hemisphere.

David P. Gaff

Findings indicated that the predicted phenomenon was: statistically
confirmed, not influenced by differences in energy levels between mask and
target stimuli, significantly affected by stimulus onset asynchrony between
target and mask, dependent on size of the mask, and not reproducible when
diffuse light was substituted for the pattern. Various aspects of these
results supported the theory that the cross-field masking effect resulted
from interaction of central (hemispheric) rather than peripheral neural
mechanisms. ‘Hethods of improving the paradigm and for testing related

theoretical issues were discussed.

"CROSS-FIELD" VISUAL MASKING BY TACHISTOSCOPIC PRESENTATION OF

TARGET AND NOISE PATTERNS To OPPOSITE CEREBRAL HEMISPHERES
By

\
David PP"‘co££

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology

1973

g.
I: a- ’9}
fly;
r 2;
e

Copyright by
DAVID PAUL.GOFF

1973

DEDICATION

To Dr. Charles Henley

ii

ACKNOWLEDGMENTS

I would like to thank Dr. Charles Henley, Committee Chairman, for
his assistance with this project and for his support as mentor, counselor,
and confidant throughout my past four years as a graduate student at
Michigan State. My appreciation also goes to Committee Members Dr. Lester
H. Hyman, Dr. Hiram E. Fitzgerald, and Dr. Robert L. Raisler for their
support. Dr. Paul Bakan has greatly contributed to the develOpment of my
interests which evolved into conducting this project, and Dr. James L.
Zacks has provided valuable technical assistance, consultation time, and
literature sources. Finally, I am indebted to my wife, Anna-Karin, for
the laborious job of typing and proof-reading this manuscript, for her
moral support, and the many personal sacrifices she has made in my behalf

over the past four years.

111

TABLE OF CONTENTS

CHAPTER 1: INTRODUCTION ............................................. 1
CHAPTER 2: EXPERIMENT I ... .......................................... 17
Method ............ ......................... . ................... 18
Results and discussion ................. . ........... . ........... 25
CHAPTER 3: EXPERIMENT II ............................................ 39
Method ............... .......... . ............ ... ......... ....... 46
Results and discussion ..... ......................... .......... . 47
CHAPTER 4: EXPERIMENT III ......... ........... . ............. .... ..... 54
Method ................... ......... ............................. 56
Results and discussion .. ..... ........... ....... ................ 59
CHAPTER 5: EXPERIMENT IV ..... 70

Method ......................................................... 73
Results and discussion .......................... ...... ......... 75
CHAPTER 6: GENERAL SUMMARY AND CONCLUSION ........................... 81
BIBLIOGRAPHY ........................................................ 88

APPENDIX A: Brief Review of Controversial Issues Concerning Left
Versus Right Field Differences in Tachistoscopically Presented
Vigual Stimli O... ...... OOOOOOOOOOOOOOO......OOOOOOOOOOOOOOOOOO 93

APPENDIX B: Errors (Guesses) Classified as "Stimulus Dependent" for
Each Target Letter Used in Experiment I (Error Analyses 1 and 2) 94

APPENDIX C: Summary Table and Analysis of Variance for Error
Analysis 1 Based on Percentage of Stimulus-Dependent Errors
‘. the Depend‘nt V‘ri.b1. O0.0.0.0.0..........OOOOOOOOOOOO...OO. 95

APPENDIX D: Summary Table and Analysis of Variance for Error
Analysis 2 Based on Percentage of Stimulus-Dependent Errors
8' the Dependent variable 00.0.0000........OOIOOOOOOOCOOOOOO0.00 96

iv

LIST OF TABLES

Table 1. Percentage of correct responses for left versus right hemis-
phere and mask versus no-mask conditions for Groups I-IV, per—
centage decrement produced by the mask, average across hemis-
phere conditions, difference in percentage decrement, and mean
target exposure time at threshold ................................

Table 2. Analysis of variance summary based on percentage correct

.core. for Experimentl O....OOOOOOOOOOOOOOOOOO00.00.000.000.......

Table 3. Percentage of stimulus-dependent errors for left versus
right hemisphere and mask versus no-mask conditions averaged for
Groups I-IV, percentage decrement produced by the mask, average
across hemisphere conditions, and difference in percentage decre-

ment (Error analysis 1) ...........................................

Table 4. Percentage of stimulus-dependent errors for left versus
right hemisphere and mask versus no-mask conditions averaged for
Groups I-IV, percentage decrement produced by the mask, average
across hemisphere conditions, and difference in percentage decre-

mnt<£rroranaly.1.2) O00............OOOOOIOOOOOOOOO......OOOOOOO

Table 5. Percentage of correct responses for left versus right hemis-
phere and mask versus no-mask conditions for Groups III and V,
percentage decrement produced by the mask, average across hemis-
phere conditions, difference in percentage decrement, and mean
target exposure time at threshold ................................

Table 6. Analysis of variance summary based on percentage correct

scores for Experiment II (Groups III and V) .......................

Table 7. Average difference in percentage correct scores, between
no-mask and mask conditions, for left and right hemisphere versus

two "stimulus-onset-asynchrony" conditions ........................

Table 8. One—tailed orthogonal comparison test of the interaction
between hemisphere and stimulus-onset-asynchrony conditions based
on the average difference in percent correct scores (no-mask minus

m.kcond1t1°n8)eeeeee ..... ......ICOOOO............OOOOOOOOOOOO0.0

Table 9. Percentage of correct responses for left and right hemisphere
conditions as a function of the no-mask, large mask, small corre-
sponding, and small non-corresponding mask conditions of Experi-
ment III (Group VI) ..............................................

25

27

33

35

47

48

49

50

S9

Table 10.

Table

11.

Analysis of variance summary based on percentage correct
scores for Experiment III .........................................

Statistical tests of the paired comparisons of all possible

combinations of masking conditions (averaged across hemispheres)

Table 12.

for Experiment III by the method of Scheffé .......................

Statistical tests of the two-way interaction effects between

left-right hemisphere conditions and all possible pairs of mask

conditions for Experiment III by the method of Scheffé ............

Table 13.

Table

Table

Table

Table

Table

Table

14.

15.

l6.

17.

18.

19.

Percent correct responses for left-right and mask - no—mask

conditions for three individual subjects receiving only the small
"corresponding" mask stimulus, percentage decrement produced by
the mask, average across hemisphere conditions, difference in per-
cent decrement, and mean target exposure time at threshold ........

Percentage of correct responses for left-right hemisphere

and no-mask - mask conditions for pattern versus light flash mask-
ing stimuli, percentage decrement produced by masking, average
across hemisphere conditions, difference in percentage decrement,
and mean target exposure time at threshold ........................

Analysis of variance summary based on percentage correct

scores for Groups V and VII of Experiment IV ......................

Summary table for error analysis 1 ..........................
Analysis of variance for error analysis 1 ...................
Summary table for error analysis 2 ..........................

Analysis of variance for error analysis 2 ...................

vi

59

6O

61

75

76

95

95

96

96

LIST OF FIGURES

Figure l. Afferent pathways of the visual system of the rhesus monkey

Figure 2. Relative strength of target and noise information arriving
in the respective hemispheres as predicted by the model ...........

Figure 3. Predicted changes in recognition accuracy for verbal target
stimuli resulting from projection of the mask pattern to the hemis-
phere Opposite that of the target .................................

Figure 4. Pathways in the visual system of the retinal images of two
objects falling on the vertical meridian, one in front of and one
behind the fixation point, for a human with sagittal section of
the Optic chiasm ..................................................

Figure 5. Example of a single neuron in the right visual cortex which
was driven by a stimulus crossing the vertical meridian for a cat
with sagittal section of the Optic chiasm .........................

Figure 6. Fixation stimulus with brackets for parallax adjustment .....
Figure 7. Example of the target, mask, and no-mask stimuli ............

Figure 8. Composite view of the stimuli as seen through the tachisto-
scope on a "masked" trial with target projecting to the right
hemi.Pher. 0.0.0.............OOOOOOOOOOOOOOOOOOO......OCOOOOOOOOOOO

Figure 9. Expected stimulus arrival asynchrony of target and mask
stimuli at the two hemispheres for onset of a left-field target
preceding onset of a right-field mask by 8 ms with an assumed
callosal transmission time of 8 ms ................................

Figure 10. Expected stimulus arrival asynchrony of target and mask
stimuli at the two hemispheres for onset of a right-field target
preceding onset of a left-field mask by 8 ms with an assumed
callosal transmission time of 8 ms ................................

Figure 11. Corresponding and non-corresponding heteronymous mask
locations for the right-field (left hemisphere) mask conditions ...

vii

4

6

7

l4

16

20

21

4O

41

56

Chapter 1: INTRODUCTION

Over the past 50 years a large volume of evidence has accrued in
support of the concept of lateral specialisation of functioning in the two
cerebral hemispheres of humans. This evidence has derived from a variety
of sources which include: direct cortical stimulation, study of patients
with unilateral brain damage, postoperative case studies, gross evoked
potential recordings, sodium amytal anesthetisation of one or the other
of the hemispheres as a preoperative investigation technique, unilateral
presentation of visual, auditory, and tactile stimuli to normal human
subjects, and most dramatic, the recent work of Sperry and associates with
"split-brain" epileptic patients. Among the most consistent generaliza-
tions made are that the left (language) hemisphere in most right-handers
specialises in symbolic and analytic functions while the right (mute)
hemisphere demonstrates a superiority in Gestalt-type processing of spatial
information (Hecaen, 1962; Gaszaniga et al., 1965; Geschwind, 1965; Levy,
1969; Levy-Agresti & Sperry, 1968; Sperry & Levy, 1970).

Nbst of the above methods for investigating asymmetry of hemispheric
functioning comprise the drawback that few brain-damaged or surgical patients
are available to the general scientific community. Furthermore, the condi-
tions leading to the use of such surgical procedures involve pre-existing
structural damage to the normal central nervous system of a rather critical
nature. The choice, therefore, of some supplementary paradigms that permit

the study of "normal" subjects seems desirable. The most popular methods

2
which fall in this category are the dichotic listening paradigm for the
auditory mode and tachistoscOpic presentation of visual stimuli to the
right or left hemisphere for the visual mode. Ximura (1961) showed a left
hemisphere superiority for verbal stimuli via the right ear, and a right
hemisphere superiority for the perception of melodies (1964). Although the
hemisphere contralateral to the stimulated ear is favored for receiving the
stimulus, each ear nevertheless has a direct cochlea-cortical projection
route to both hemispheres (Davis, 1951), a "disadvantage" not shared by
the visual system. For higher mammals, the two right hemiretinae (left
visual field) contain direct retina-cortical projections to the right
hemisphere only while the two left hemiretinae project directly to the
left visual cortex (see Thompson, 1967, p. 250). Thus visual stimuli can
be projected, at least initially, to only one hemisphere by tachistoscopic
exposure to the contralateral (Opposite) visual field.

The vision literature on laterality is extensive and not without
controversy (see review by White, 1969). The general findings, however,
support a left hemisphere superiority for verbal and a right hemisphere
superiority for spatial processing as indicated by more accurate recogni-
tion or shorter reaction times to tachistoscOpic presentation of the
appropriate material in the appropriate visual half-field (Ximura, 1966;
White, 1969; Rissolatti et al., 1971)1.

While some of these vision studies show significant differences in
support of the lateralisation of function hypothesis, the differences are
nevertheless quite small and many of the studies are either contradictory
or totally inconclusive. The most outstanding difficulty in this approach

with normal subjects is, of course, the fact that 6-10 ms (Efron, 1963a;

 

lThis literature has been reviewed and the troublesome issues dis-
cussed in Appendix A.

3
Jeeves & Dickson, l970)after one hemisphere has been stimulated, the in-
formation has become available for processing by the Opposite hemisphere
via the corpus callosum. The function of the callosum in this transfer of
visual information has been demonstrated with commissurotomized animal and
human patients (Sperry, 1961; Geschwind, 1965). ElectrOphysiological
evidence has been provided by Berlucchi et a1. (1967). Hence, either or
some combination of both hemispheres may still do the processing.

Clearly, if one is to investigate more complex modes of unilateral
"information processing" in "normal" human subjects, a method is needed for
isolating, as much as possible, the processing of Specific information to
a given hemiSphere. One means by which this may be possible would be to
present target information to one hemisphere while presenting "conflicting”
or noise pattern stimuli to the contralateral side and perhaps interfere
with its processing. The specific aim of this project has been to develOp
and test the feasibility of such a model for the visual system. The empir-
ical data will provide some tests of the hypothetico-deductive assumptions
on which the model is based. Following is a detailed description of the
model along with the behavioral, anatomical, and neuTOphysiological evidence

which seem to support its tenability.

grief anatomy.

Figure l is a block diagram of the afferent pathways of the visual
system of the rhesus monkey which relate specifically to the processing of
patterned stimuli. For simplicity, only the pathways from right hemiretinae
corresponding to the left field of vision have been depicted. In primates,
the retina-cortical projection (for the contralateral half-field of vision)
goes exclusively to area 17 (Wilson a Cragg, 1967) of the visual cortex.
Area 17 then projects to 18 and 19 on both sides (Cragg et al., 1969;

Zeki, 1969; Zeki, 1970); 18 and 19 from both sides then project to the

Eight hemiretinae

temporal nasal

I
I
‘t. Optic ,
\ chiasm ’
\ I
\ I
\> I
’I

lateral geniculate
.2, ® nuc leus
Z////////””—_- anterior _—~§‘\\\\\\\\\\5
cogeismtzf

 

 

 

 

 

    

 

 

 

 

 

 

 

[inf::::::poral __IL-"TL:_¢7 ‘ \ f I
\w /‘i /1 19/
2:32;: \“’ ><\ 718/

 

 

 

 

4
L _____ hl/
spleneum of 17

corpus
callosum

Figure l. Afferent pathways of the visual system of the rhesus monkey (most
like men). Connections to structures (e.g. pulvinar and superior colliculus)
not related to pattern processing are excluded. (Drawn from a description
by Gross, et. al., 1972)

S
inferotemporal cortex (Cragg et al., 1969). All the cortico-cortical path-
ways just described go through the spleneum of the corpus callosum. The
two inferotemporal cortices connect through the anterior commissure (Fox
et al., 1948). Present evidence suggests that these pathways from area to
area of a given side represent successive stages of pattern processing

(Hubel & Wiesel, 1965; Gross et al., 1972).

The model

Given that the two eyes are focused on a fixation point, a "target"
stimulus tachistoscopically flashed - for a time shorter than the 200 ms
central reaction time for saccadic eye movements (Robinson, 1968) - to, say,
the left half-field of vision, will go to area 17, the "primary" receiving
area, in the right visual cortex (refer to Figure 1). The information will
then go to the "secondary" receiving areas 18 and 19 on the ipsilateral side
and to the secondary areas via the longer transcallosal pathway on the
contralateral side. The writer proposes that the "raw" stimulus informa-
tion, coded in electrophysiological form, will be more attenuated when it
arrives at the contralateral as compared to the ipsilateral secondary re-
ceiving area as a result of the longer transcallosal pathway. (These
assumptions will be discussed below.) Considering, now, the other half of
the system, a visual "noise" pattern simultaneously flashed in the right
visual field will arrive in the secondary areas with greater intensity in
the left as compared to the right hemisphere. The result, for this partic-
ular example, would be that the right hemisphere receives a strong target
signal (compared to the left) and a weak noise pattern (compared to the
left). Correspondingly, the left side receives a weaker target signal and
a stronger noise pattern (see Figure 2). Coined in different terms a high

"signal-to-noise" ratio is obtained on the right side, and a low

6
signal-to-noise ratio on the left. If the target and visual noise pattern
are located in corresponding heteronymous (mirror symmetric) areas of the
half visual fields, and not too far from the vertical meridian (less than
7° visual angle), they should project to a common location in visual area
18 (see for example Berlucchi & Rizzolatti, 1968; and Blakemore, 1970).
It is proposed that the noise pattern will interfere with (mask) the target
stimulus in such a way as to decrease recognition accuracy for the target.
This "masking" effect should occur on both sides (since both stimuli go to
both sides), but it should be greater on the side with the lowest signal-
to-noise ratio, i.e. the left hemisphere for this particular example.

A G)

I \
l \
I \

 

 

 

I, \
(.1 L)

weak strong

target target
1

strong weak

noise noise

1
left right
hemisphere hemisphere

Figure 2. Relative strength of target and noise information arriving in
the respective hemispheres as predicted by the model.

Employing, now, the previously described concepts of lateral
specialisation of function of the two hemispheres, it should be a simple
matter to derive the behavioral predictions generated by the model. Take,
for example, the left (language) hemisphere as the required processor. It
is a given at this point that a verbally coded response to verbal stimulus

material requires that the information reach the left hemisphere.

7

A stimulus (letter or word) flashed to the right hemisphere must cross the
transcallosal pathway, and the signal attenuation hypothesis predicts that
recognition accuracy should be less than when the stimulus is flashed

directly to the left hemisphere (see Figure 3. no mask condition). These

are the standard observations as described in the above studies. With the

recognition

X
accuracy I
k

 

 

 

tar'get tar'get
to left h to right h

Figure 3. Predicted changes in recognition accuracy for verbal target
stimuli resulting from projection of the mask pattern to the hemisphere
opposite that of the target. (The finding Xj>k would satisfy the model)

target flashed to the right and noise pattern now flashed to the left hemis-
phere, recognition accuracy should decrease by some amount X, as compared

to the right hemispher "no mask" condition (Figure 3). The responding

 

hemisphere now has the low signal-to-noise ratio. To assess the undesirable
(for present purposes) effect of the mask projecting to the opposite hemis-
phere via the cortico-cortical pathway, we reverse (laterally) the above
stimuli and flash the target to the left hemisphere and the noise pattern

to the right. Recognition accuracy should now decrease by some smaller

amount k (from left hemisphere - no mask condition), since the left or

8
responding hemiSphere now has the larger signal/noise ratio. Thus, if the
X:>k prediction is the empirical finding. then the data will support the
model. Secondly, and as an algebraic consequence of the above statements,
the difference in recognition accuracy between left and right hemisphere
presentations should increase under the masking condition, and the magni-
tude of that increase will determine the utility of the paradigm as a re-

search tool.

Qichoptic visual masking

 

That interference between a target and noise pattern projected by
separate pathways can be produced at the cortical level receives strong sup-
port from the dichOptic visual masking studies. Schiller (1965) flashed a
letter to one eye and either a flash of light or a visual noise pattern to
the other at corresponding retinal areas (foveally) and found a significant
increase in recognition threshold (in ms exposure time) above the no-
masking condition for the noise pattern - but not the diffuse light. These
results indicate an interaction between stimuli presented to the two eyes at
some central location subserving a binocular interaction between the two
eyes. Area 17, which receives a direct retina-cortical projection from both
eyeg.serves as a good candidate.1 Hubel and Wiesel (1965, cat; 1968, monkey)
find that many "simple" cells in area 17 can be binocularly driven from
corresponding homonymous points of the two retinae and that they respond

only to lines and edges (pattern fragments), not diffuse light.

 

1In their spatial vision review article, BishOp and Henry (1971, p. 145)
state: "Relatively little interaction between the pathways from the two eyes
takes place below the level of the striate cortex (Sanderson et al., 1969;
Singer, 1970)." The authors they cite, of whom Bishop is one, found that
some inhibitory binocular interaction does occur in the lateral geniculate
body of the cat. The relevant pathways are unknown and no speculation has
been made concerning their purpose. It should also be noted that the inter-
action is (a) extremely small and (b) exists only for corresponding homonymous
points.

9

Referring to Figure 1, we see that the difference between the
dichOptic and "cross-field" masking paradigms is as follows: in the
dichoptic case, the stimuli may interact in area 17 which receives direct
retino-cortical projections from homonymous retinal areas; in the "cross-
field" case interaction cannot occur until area 18 which receives projec-
tions from ipsilateral 17 (right hemiretinae in the diagram) and contra-
lateral 17 (left hemiretinae) via the transcallosal route. In the latter
instance, individual cells in area 18 would have to be driven from corre—

sponding heteronymous points in the two half-fields of vision. Direct

 

evidence for this latter possibility has recently been obtained by
Berlucchi & Rizzolatti (1968) who found neurons in area 18 that could be
binocularly driven from corresponding heteronymous areas in the split-

chiasm cat (Figure 5).

Some functions of the corpus callosum

This area is too extensive to receive a detailed review here (see
Ettlinger, 1965; BishOp & Henry, 1971); however, a few points relevant to
the theoretical development of the above model require discussion. They
concern the nature of information transfer between the visual cortices
of the two hemispheres and its relation to the issues of interhemispheric
masking. The following introductory paragraph will facilitate the discus-
sion of these points.

Hubel and Wiesel (1965), using the retrograde degeneration technique,
found that fibers going from area 17 to ipsilateral and contralateral l8
and 19 projected to very specific areas; the area in 18 immediately adjacent
to the 17-18 border, and the area in 19 immediately adjacent to the 18-19
border. The retinotopic projection to these areas represents the areas

adjacent to the vertical midline of the field of vision. For example,

10
they found (by microelectrode recording; Hubel & Wiesel, 1967) that the
"receptive fields" of all cells on the 18 side Of the 17-18 border ex-
tended to the vertical meridian. Secondly, Berlucchi and Rizzolatti
(1968) found, as mentioned previously, that cortical cells in this same

section of area 18 could be binocularly driven from correSponding heter-

 

onymous retinal areas adjacent to the vertical midline. This evidence
suggests that two of the "natural" functions of this callosal crossover
network may be to (a) provide continuity between the two half-fields of
vision (Hubel & Wiesel, 1967; Berlucchi & Rizzolatti, 1967) and (b) pro-
vide a stereOptic mechanism for making depth judgments about objects
falling on the vertical meridian and thus projecting to Opposite hemis-
pheres (see discussion and review by Bishop & Henry, 1971). It is by this
particular crossover network through the spleneum of the corpus callosum
that the prOposed interference phenomenon between the hemispheres might
occur.

One feature of the cross-field masking model described above was
the prOposal that interference might be produced in the 18-19 areas by a
visual noise pattern. This presupposes that, since a noise pattern is to
be the effective masking stimulus, it must be interfering with "basic
sensory data" (e.g. pattern fragments) of the target stimulus projecting
from contralateral 17 rather than some more highly processed code for the
stimulus (e.g. in engram form). Is there any evidence to support the con-
tention that the spleneum transmits basic sensory data as Opposed to more
complex codes? Myers (1962) and Sperry (1964) argue that this is, in fact,
the case but their support is based on some rather indirect inferences
deriving from the interhemispheric transfer of pattern discrimination learn-

ing in chiasm-sectioned animals. Far more direct evidence was provided by

ll
Berlucchi and Rizzolatti (1967). In an effort to specifically answer the
above question they conducted a microelectrode recording analysis Of the

receptive fields of the axon fibers in the spleneum (posterior third) of

 

the corpus callosum of the cat. They found all the fields to have pro-
perties identical to those of the simple, complex, and hypercomplex cells
found by Hubel and Wiesel (1965) in areas 17, 18, and 19 of the cat. The
fields extended from the vertical midline to as far as 200 into the peri-
phery.

While the above obviously supports the hypothesis that the Spleneum
may be specialized for the interhemiSpheric transfer of sensory-type visual
information, it does not exclude the possibility that yet another commis~
sural structure may transmit more highly coded information between "higher"
visual centers of the two hemispheres - in particular, the anterior commis-
sure which links the two temporal lobes. The evidence on this point is
short but rather dramatic. Gorden, Bogen, and Sperry (1971) describe the
testing of two human patients having received partial commissurotomy Opera-
tions for the control of epilepsy. Both had only the anterior commissure
sectioned with all other structures remaining intact. These patients could
easily give a verbal description of visual stimuli flashed to the right
hemisphere. However, a third patient (described in the same report) who
had received only a sectioning of the spleneum - all other structures re-
maining intact - was totally unable to describe any visual material flashed
to the right hemisphere. In sum, it appears that interhemispheric communica-

tion of visual information depends on the existence of an intact spleneum

 

which has been shown, at least in a lower species, to carry only sensory-
type codes.

The last events in the processing chain that would lead to the

Observation of a cross-field masking phenomenon have involved the implicit

12
assumption that interference with the processing of target information in
areas 18 and 19 would necessarily disrupt "higher" processing in the visual

centers of the temporal lobe and therefore the output in response form

 

produced by the reSponding hemisphere. Support for this contention derives
from a series of cross-lesion experiments on monkeys performed by Mishken
(1966). He demonstrated rather conclusively that the contribution of the

inferotemporal cortex to the pattern recognition process depends on the

 

neural processing that occurs in areas 18 and 19 (see also Stone & Freeman,

1973).
While the above argument makes a case for the hypothesis that mask-

ing in areas 18 and 19 could disrupt processing and output by higher centers,

 

it does not exclude the mutually igclusive possibility that yet another
pathway, the projection of 18 and 19 to the inferotemporal cortex via the
spleneum could provide the target information for continued processing by
the "masked hemisphere" (refer to Figure 1). Although the visual noise
pattern also projects to this site from the ipsilateral side, the above
suggests that masking, in a sense. "engram with engram" could yet be a more
effective procedure than masking "engram" with "noise."1 If this were
tenable, then masking, for example. a target letter or word with a matrix
of random letters would have relatively EBEE effect than masking with a
visual noise pattern. A testable hypothesis.

Finally, it was mentioned previously that a "natural" callosal func-

tion which could potentially account for an interhemispheric interaction

 

1 For lack of a better word, the term "engram" is used, not in its
strictly conventional meaning as a stored bit as a result of learning, but
to mean the terminal stage to which a given pattern is processed by the
visual system. The word "code" also seems inapprOpriate since the "basic
sensory information" is obviously in coded form even in the peripheral
energy transducers.

 

13
between target and mask, was the mechanism providing for a stereOptic inter-
action for stimuli that projected binocularly to Opposite hemispheres (a
situation that is identically simulated by the cross-field visual masking
paradigm). As much of the above data are based on cats and monkeys, and
since, as we ascend the phylogenetic scale, the visual system becomes more
precise (e.g. receptive fields and nasotemporal overlap are smaller for
monkey than for cat etc.), the question arises as to how far into the peri-
phery, on opposite sides of the fixation point, can we safely assume a

binocular interaction from heteronymous points in the human subject. The

 

data are provided by the unfortunate case of a 14-year-old boy who fell Off
his bike and Split his Optic chiasm (Blakemore, 1970). The subject was able
to make depth judgments when a slit was briefly exposed in front of his fixa-
tion point although its images projected to the temporal retinae of both

eyes and therefore projected separately to the two hemispheres (see Figure 4).
The amount of disparity that could be tolerated and still give the stereo-
sCOpic effect was 6°, i.e. 3° to the temporal side of each fovea. This

study provides not only behavioral evidence for an interhemispheric pathway

 

for binocular interaction in man (which correlates with the electrophysio-
logical findings of Berlucchi and Rizzolatti, 1968), but an assessment of

the spatial parameters relevant to the present project.1

The signal attenuation hypothesis
Attenuation of signal and mask due to the longer transcallosal route

is perhaps the most critical assumption on which this model is based. Three

 

1The question arises as to the possible contribution of the "naso-
temporal overlap" (the small band at the fovea from which fibers might pass
to either Optic tract) which was found to be 0.9° wide in the cat (Stone,
1966). In man and monkey it is purported to be considerable smaller (van
Buren, 1963). Functionally, it does not seem to be an important factor in
man since Mitchell and Blakemore (1970) found that section of the corpus

callosum in man (chiasm intact), "disrupts depth perception entirely in the
middle of the visual field."

 

 

 

 

 

 

 

lateral ‘ '
geniculate 13 43
nucleus ‘ I I Q
I I
I
i :
I
I ..---.u. '
. I I '
. I ' I
I I I I
I I I I
I I . .
A : ' A
O)--—1---:----O 4: ‘visual
I.) 1I I 4) cortex
A I I
mm. L---
corpus 4
callosum

Figure 4. Pathways in the visual system of the retinal images of two objects
falling on the vertical meridian, one in front of and one behind the fixation
point, for a human with sagittal section of the Optic chiasm. The Object at
A, directly in front of fixation point F, projects to the two temporal hemi-
retinae and drive the neurons indicated by the solid circles. The neuron at
A, left hemisphere, receives an input from corresponding heteronymous areas
Of the two eyes, one input deriving from the corpus callosum. Object B can-
not be seen due to interruption of decussating fibers at the Optic chiasm.
(figure redrawn from Blakemore, 1970)

15
kinds of evidence will be offered for its support.

Bremer (1965) electrically stimulated the lgfg lateral geniculate
body (see Figure 1) of the brain of a cat and measured the gross evoked
potential responses from symmetrical areas of the left and right visual
cortex. The oscilloscOpe traces Obtained from the lgfg hemisphere re-
presented the direct geniculo-cortical pathway while those obtained from
the E£EE£ hemisphere represented the geniculo-cortico-cortical pathway via
the corpus callosum (Ibid. 1965, p. 286). The two wave forms were not
only similar, as might be expected, but the amplitude of the signal from
the sigh; hemisphere (longer pathway) had decreased by a factor of about
five when compared to the left hemisphere (direct pathway).

Behavioral evidence of the "inefficiency" of the callosal trans-
fer mechanism comes from studies by Myers (1962). He found that split-
chiasm animals could learn a pattern discrimination task through one eye
(while the other is occluded) and then perform the learned task using the
Opposite eye even after post-training sectioning of the corpus callosum.

In other words, both hemispheres learned the discrimination, one of them
through the callosum. However, the "untrained hemisphere" always performed
more poorly than the one which had received the discriminanda via the direct
retina-cortical route.

At the single cell level, Berlucchi and Rizsolatti (1968) provide
the most dramatic evidence. These researchers found that cells in area 18
of the chiasm-sectioned cat (see Figure 5) could be binocularly driven from
corresponding heteronymous retinal areas (as previously described). Further-
more, a cell when stimulated (by natural visual stimuli) via the callosal
pathway gave responses that "were in general less brisk and more fatigable"

than when stimulated by equal stimuli via the direct retinO-cortical path.

 

16

left

 

 

 

 

 

 

 

lateral

geniculate

nucleus
r""“l
| I
I I
I I
g I
I I
I I
I I
I I
I I
I T
L-_-J
corpus

.callosum

 

microelectrode

Figure 5. Example Of a single neuron in the right visual cortex which was
driven (Berlucchi & Riszolatti, 1968) by a stimulus crossing the vertical
meridian for a cat with sagittal section of the Optic chiasm. The recorded
cell could be driven either by the part of the stimulus falling in the left
visual field (white area) and projecting to the right temporal hemiretina
or by the stimulus in the right visual field projecting to the left temporal
hemiretina (black area) and across the corpus callosum. The line at A in-
dicates the division between the nasal and temporal hemiretina of each eye.
(figure redrawn from Berlucchi and Rissolatti, 1968)

Compare Figures 4 and 5 and note that similar events in the visual
system are produced by two different stimulus configurations. Figure 4
applies to the case that would represent binocular stereOpsis and Figure 5

represents the situation encountered in the cross-field visual masking
paradigm.

Chapter 2: EXPERIMENT I

For all studies reported in this paper, the "target" stimuli em-
ployed were single capital letters for which the left (language) hemisphere
is presumed to be the primary, or final, processor for naming the letters
and initiating a verbal response. To briefly reiterate the basic paradigm,
single letters are to be flashed unilaterally and in random sequence to
both left and right visual hemifields under both no-mask and mask (noise
pattern in hemifield Opposite the letter) conditions. The main prediction
is that with target letter projected to the right and noise pattern to the
left hemisphere, decrement in response accuracy should be greater - com-
pared tO the no-mask condition - than when the situation is reversed, i.e.
letter to left and mask to right hemisphere (Figure 3). The former is pur-
ported to represent the low "signal-to-noise" and the latter the high
"signal-to-noise" stimulus conditions for the lg££_hemisphere.

The results of exploratory pilot work supported, though not con-
clusive statistically, the above predictions and provided information for
establishing the base-line parameters to be used for the present studies.
The goals of the experiments reported here were (a) to establish, if possible,
statistical confirmation of the predicted "effect," (b) to provide a system-
atic investigation of some stimulus parameters which may be relevant to in-
creasing or decreasing the "effect," and (c) to test some related theoretical
assumptions about the neural mechanisms potentially involved.

The purposes of Experiment I were (1) to accumulate sufficient data
to make a statistical determination of the tenability of the "effect" and
(2) to investigate the possible effects Of varying brightness relationships
between target and noise pattern. Brightness was varied by holding target

17

l8
intensity constant for different noise-pattern intensities and by holding
noise intensity constant while varying target intensity. The preliminary
supposition was that, as with other masking paradigms, the greater the
brightness difference between target and mask, the greater would be the
masking effect. In the present case, however, the issue becomes consider-
ably complicated by the fact that the (presumed) masking site is exclusively
cortical. The present goal was to find a set of relative brightness para-
meters which might suggest the existence of an Optimum set for obtaining

the differential effect.

METHOD

Subjects

The subjects were 16 paid volunteers from two undergraduate
psychology classes at Michigan State University. All were males, right-
handed, as determined by the Crovitz and Zener (1962) questionnaire for

hand dominance, and naive to the purpose of the experiment.

Apparatus
A three-channel Scientific Prototype (Model GB) Auto-TachistoscOpe

was used. Each field was back-lit through the diffusion screen (with
color-matching filters) by the standard fluorescent bulbs (GE-FTA-S with
painted silver stripe between the electrodes). Although not measured,

the rise and decay times are claimed to be considerably less than 1 ms

each. Each of the two "stimulus" fields were modified to obtain more equal
left-right half-field brightness by repositioning the right member of the
vertically mounted pair of bulbs and by inserting a mat white strip of
cardboard behind each bulb pair. Maximum and minimum stimulus field bright-
Iaesses, measured through the split-beam binocular eyepiece with a Salford

lilectrical Instruments Company spot-photometer, were 2.50 footlamberts (ft.l.)

l9
and 0.40 ft.l. respectively. For two Of the fields, stimuli were presented

on 24x36 umlslide transparencies automatically fed from lOO-slide "rototrays."

Stimulus materials

The fixation stimulus viewed by the subject was a dark field, and
a lighted circle in the center (0.44° visual angle in diameter) within
which was an "x" containing a "dot" (0.088° visual angle) in its center
(Figure 6). The subject also saw a small lighted bracket in each of the
four corners of the rectangular field. The area within the brackets re-

presents the unoccluded portion of any stimulus field as seen through the

 

Figure 6. Fixation stimulus with brackets for parallax adjustment.

binocular eyepiece and permitted the subject to make minor self-centering
head adjustments before each exposure sequence. The magnified (through
the scope Optical system) portion of this unoccluded viewing area had an
"apparent" viewing distance of 30 inches and projected a visual angle of
2.9°x5.25° (laterally) at the retina. The lighted areas (center circle
and brackets) Of the fixation field measured 0.40 ft.l. while the "black"
area of the transparency (found under higher luminance conditions to trans-
mit 0.81 Of the back light) was 0.0032 ft.l.

The target stimuli (see example in Figure 7) consisted of a rototray

Of 100 slide transparencies, each with a single black letter Of the alphabet

20

against a "white" background on the left or right half of the slide while
the half-aide Opposite the letter was "black". The center point of each
letter was 1.70 to the left or right of fixation as viewed through the
scope. The letters subtended an angle of 0.340 in height and a maximum

of 0.340 in width. Thickness of the letter features was 0.030 visual angle.

 

   

 

L______-

target stimulus mask stimulus

 

no—mask stimulus

Figure 7. Example of the target, mask, and no-mask stimuli.

A second rototray, for the remaining stimulus channel, contained a
set of 50 blank slides, blank the entire area, and a set of 50 masking
stimuli (Figure 7) which were black on either the left or right side and
contained a visual noise pattern, the area of which was 501 white and 50%
black, on the Opposite side. This pattern is identical to that used by
Schiller (1965) for dichoptic masking Of single letters. A third rototray
identical to the second except for a one to one reversal in order of mask

and blank stimuli was also used.

21

All stimuli were prepared in black-white reverse on mat white paper with
black marking pen and photographed with Kodak High Contrast Copy Film 5069.
The negatives then comprised the stimulus material. The black areas of
the mask and blank slides were exposed and develOped to the same density
so that they would pass an equal amount of light from the rear diffusion
screen in their specific channel.1

In the center of the target and masking fields, a black vertical
strip, projecting a visual angle of 1.00 at the eyepiece, was mounted
immediately in front of the stimulus position. This strip was to block
foveal transmission from either half-field resulting from whatever naso-
temporal overlap exists in man (estimated to be considerably less than
the 0.90 found in the cat; van Buren, 1963), or from microsaccadic eye
movements, 7 to 10 minutes to either side of fixation (Ditchborn & Gins-
burg, 1953; St. Cyr & Fender, 1969). The resultant composite viewed by
the subject on a "mask" trial and target presentation on, say, the left

side is illustrated in Figure 8. The remaining unoccluded portions of the

 

Figure 8. Composite view of the stimuli as seen thrOugh the tachistoscOpe
On a "masked" trial with target projecting to the right hemisphere.

 

LThis was crucial for the purpose of maintaining a constant back-
ground illumination in the half-field of the superimposed target letter
for both mask and no-mask trials.

22
mask projected a visual angle of 2.130 in width and 2.900 in height. A no-
mask trial appears identical except for replacement of the noise pattern by

a "black" half-field.

Procedure

As the subject sat in a chair facing the tachistoscOpe, a head rest,
chin rest, and the interocular separation of the eyepieces were all adjusted
so that the subject, by occluding first one eye, then the other, could see
all four brackets of the fixation field with each eye separately. The
room lights were then turned off and the subject was given a brief period
to dark-adapt to the brightness level of the fixation field.

At the beginning of each trial, the subject binocularly viewed
the fixation slide. At the signal "ready", the subject made minor head
adjustments (if necessary) to insure a clear view of all four reference
brackets. About 1% seconds after the ready signal, the subject was
signalled to "focus" on the center dot, and within another second the
stimulus fields were fired by the experimenter.

On a "masking" trial, the noise pattern fired on one side for a
total duration of 200 ms. Fifty ms after onset of the noise pattern the
target flashed on the Opposite side for a duration of 3.5 to 9.0 ms depend-
ing on brightness conditions and threshold level for a particular subject.
On a "no mask" trial, all conditions were the same except for replacement
of the noise pattern with an all "black" slide (i.e. the masking field
still fired for the same time duration to maintain constant target back-
ground for 2252 "mask" and "no mask" trials). The fixation stimulus
remained continuously lit. At the termination of a 750 ms interval (from
trial onset), the slide changers automatically advanced the target and

Inask - no-mask stimuli for the next trial. The subject's task was to

23
call out a letter as quickly as possible and within the 750 ms interval,
i.e. before he heard the slide changers advance. The subject's response
was manually recorded during an intertrial interval of approximately
4 seconds. Three aspects of the experiment were emphasized to the sub-
ject as having the following order of importance: (1) he must constantly
maintain the fixation point for the period beginning with the "focus"
command and ending with the sound of the slide changers, (2) for each
trial, he was to make his best guess as to what the letter might have
been, regardless of the stimulus conditions and regardless of whether
or not he actually "saw" a letter, and (3) he was to call out his best
guess before he heard the changers advance.

The target set Of stimuli contained 25 letters of the alphabet
(excluding Q), each letter appearing 4 times, twice on the right and
twice on the left. The order of the letters and their right-left posi-
tion was randomized with the restriction that the same letter did not
appear more than twice in succession, and not more than three right or
three left presentations occured in succession. The "mask - no mask"
stimulus set was randomly arranged with the restrictions that (a) each
letter was "masked" once on each side, and (b) no more than three mask or
three nO-mask conditions occurred in succession. The second "mask -
no mask" set (100) was identical to the first except for a one to one
reversal of the 50 noise and 50 blank stimuli. In other words, a letter
in a given serial position that was masked on the first set of 100 trials,
was not masked the second time around. In sum, the right-left, mask -
no mask, and serial position of the mask - no mask conditions were counter-
balanced for each letter, with conditions and letters Otherwise occurring

in random sequence.

24

At the beginning of the first session, 50 to 75 practice trials
were given to acclimate the subject and to determine the target eXposure
time which elicited a probability of correct responding at approximately
0.5 averaged across all conditions. After a brief rest period, 100 test
trials were administered with the first set of "mask - no mask" stimuli.
Occasionally, the exposure times were varied by 5% increments to maintain
the threshold responding level. Following a 5-10 minute rest period,
another 100 trials were given with the second set of "mask - no mask"
stimuli and rototrays Operating in the reverse direction. This ended the
first session of 200 test trials. A second session was given within
3-4 days (but never on the same day) with the direction of the rototray
for a particular set of "mask - no mask" stimuli reversed from what it
‘was the first testing day. A 30-40 trial warm-up was required to obtain
a threshold level of responding at the previously determined target ex-
posure time. For gpph of the four 100 trial sets, the stimulus rototray
‘was started from a random position. Each of the two daily sessions lasted
approximately 1% hours.

The 16 subjects were randomly assigned to four groups with 4 sub-
jects in each group. The groups (conditions) varied with respect to the
relative brightness difference between the noise pattern and target stimulus
fields as follows. Group I: masking field - 2.50 ft.l., target field -
0.40 ft.l.; Group II: mask - 1.05 ft.l., target - 0.40 ft.l.; Group III:
mask - 2.50 ft.l., target - 0.20 ft.l.; Group IV: mask - 2.50 ft.l.,
target - 0.10 ft.l. For Groups I and II, target brightness was held
constant while mask brightness was decreased by 501 of the difference
between the two. For conditions I, III, and IV mask brightness was held
constant while target brightness was decreased in 50% increments from
0.40 to 0.10 ft.l. by Kodak neutral density filters 0.30 (50% transmission)

and 0.60 (25% transmission) respectively.

25

RESULTS AND DISCUSSION

Percent correct, based on 100 Observations per condition per sub-

ject, was averaged across the four subjects in each group. These data

are tabulated for each of the four groups in Table 1 (within the rec~

tangles). A comparison of the mask versus nO-mask conditions averaged

across hemispheres is made in column 3. For each group, the third row

Table 1. Percentage of correct responses for left versus right hemisphere
and mask versus no-mask conditions for Groups I-IV, percentage decrement

produced by the mask, average across hemisphere conditions, difference in

percentage decrement, and mean target exposure time at threshold.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

diff. in mean T exp.
T to T to average I decr. time (ms) @
left H right H across H RH-LH threshold
Group I no-mask 63.8 59.0 61.4
til-2.50 ft.l. mask, 43.3 34.3 38.8 9.7 4.9
'T-0.40 ft.l. Z decr. 32.2 41.9 37.0
Group II no-mask 56.8 55.8 56.3
bi-l.05 ft.l. mask 46.8 40.8 43.8 9.3 4.4
'T-0.4O ft.l. Z decr. 17.6 26.9 22.3
Group III nO-mask 67.5 57.0 62.3
III-2.50 ft.l. mask 47.0 34.8 40.9 8.6 8.3
‘T-0.20 ft.1. 2 decr. 30.4 39.0 34.7
Group IV no-mask 64.5 53.0 58.8
‘M-2.50 ft.l. mask 49.0 34.8 41.9 10.4 10.6
T-0.10 ft.l. 2 door. 24.0 34.4 29.2
Groups I—IV no-mask 63.1 56.2 59.7
average mask 46.5 36.1 41.3 9.4
1 decr. 26.4 35.8 31.1
Note. H-mask; T-target; H-hemisphere; Rsright; L-left; Zdecr.=percentage

decrement produced by mask

in the table shows the percentage decrement in responding level produced

by masking - using the no-mask condition as baseline responding level.

The fourth column shows the difference in percent decrement, right minus

26
left hemisphere, and the fifth column shows the average target exposure
time required to maintain threshold level of responding. A "one between -
and two within - subjects variables" analysis of variance (see Myers, 1967,
pp. 198-209) is summarised in Table 2.

Across hemispheres and groups, the average percent correct for the
no-mask condition was 59.71 and for the mask condition 41.32, a decrement
from the no-mask baseline of 31.11 (p410.0005). The non-significant AC
interaction term indicates that the magnitude of the over-all masking
effect did not differ for Groups I-IV, the different mask-to-target bright-
ness ratios. This point will be discussed subsequently.

Although not of primary interest to the purpose of this study,
the usual left hemisphere (right field) superiority for recognition of
alphabetic material was Observed (p410.005). The non-significant AB
interaction term indicates that this effect did not differ for Groups I-IV.
Normally, such simple alphabetic material as single alphabetic letters
presented over periods of time which afford extended practice (as in this
study) shows considerably smaller left-right effects than more complex
stimuli such as words (left hemisphere) and spatial stimuli (right hemis-
phere) presented less frequently. Some of the pilot subjects receiving
400-800 presentations and following experiments employing a smaller number
of subjects did not show a significant over-all left hemisphere superiority.

The crucial interaction, BC (Table 2), hypothesized from the model,
was significant in the direction predicted (p410.05). The masking decre-
ment, averaged across groups, for the left hemisphere (target to left h,
mask to right h) was 26.41 below the no-mask baseline while a 35.82 de-
crease was obtained for the right hemisphere (target to right h, mask to

left h). The former represents the high and the latter the low

Table 2.
for Experiment I

Source

Total

Between subjects (8)
Groups41A)

S/A

Within subjects
Hemispheres (B)

A!

SB/A

Mask conditions (C)
59

sc/A

g9

£59

SIC/A

 

*p<0.05 (one-tailed);

U

12
1
3

12

**p<0.005 (one-tailed);

27

Analysis of variance summary

SS

9257.980
541.230
25.668
515.563
8716.750
1198.886
219.801
871.563
5383.886
255.301
570.063
47.270
6.418
163.563

 

1Since variances decrease as proportions deviate from 0.5, the

la

8.556
42.964

1198.886
73.267
72.630

5383.886
85.100
47.505
47.270

2.139
13.630

I'II

0.119 (ns)

16.507

1.009 (ns)

113.333
1.791 (ns)

3.468
0.157 (ns)

based on percentage correct scoresl, 2

ln

4.063**

10.646***

1.862*

***
p<0.0005 (one—tailed)

analysis should more appropriately be made on the arc sin transforms of the
Such analyses were
conducted on the transformed data for this and succeeding experiments and
led to no difference in conclusions drawn or significance levels reported

observed proportions (tabulated here as percentages).

herein.

are based on the data in their originally observed form.
2The one-tailed t

hypotheses tested in this
dictions derived from the
left hemisphere should be

right, and masking should
one of all possible within-subjects interactions had been predicted, i.e.
that masking should be greater when targets are projected to the right hemis-
The apprOpriate analysis of this interaction
for this model (Myers, pp. 337-345) is a one—tailed test of the apprOpriate

phere as compared to the left.

orthogonal comparison of the four interaction means.

produce a decrement.

For ease of interpretation, then, all statistical analyses reported

tests are reported for the following reason: all
analysis are based on "one-tailed" (a priori) pre-
model, e.g. targets projected initially to the

more accurately reported than when projected to the
Most important, one and only

Since that test as

well as the preceding two are based on one degree of freedom, they should
be algebraically identical to taking the square root of P and doing a one-

tailed t test.

and the orthogonal comparisons and producing identical results.

This was confirmed for these data by doing both the anova
(See Hays,

1963, Ch. 14, for a more thorough discussion of orthogonal comparisons for

interaction terms.)

28

"signal-to-noise conditions" for the left hemisphere. The average differ-
ence in percent decrement, right minus left hemisphere, was 9.42. This
difference, a single index of the direction and magnitude of the inter-
action, is tabulated in column 4, Table l, for each of the groups. This
column, and the non-significant interaction between groups (A) and the
within-subjects (BC) interaction, i.e. A(BC), shows that the hemisphere)(
mask interaction did not differ with varying target-to-mask brightness ratios.

The significant hemispheres x masking interaction effect, and its
prediction in advance of the empirical observation, supports the major con-
tention of the model develOped herein, that visual masking may occur inter-
hemispherically as well as intrahemispherically (the dichOptic paradigm).
However, the small magnitude of the effect, as observed so far, would seem
to preclude its utility as a general research tool for the investigation

of unilateral information processing. Each group taken separately (n-4)

 

and all but two individual subjects1 failed to produce a statistically
significant interaction. It should also be noted here that two of the
16 subjects in this sample (12.51) produced a non-significant interaction
in the direction apposite to that predicted by the model. This is in
close agreement with the observation (Hilner, et al., 1964; Kimura, 1961)
that approximately 101 of right-handed males have the language center in
the right hemisphere as determined by unilateral sodium amytal injection
into the left or right carotid artery.

The failure to find that varying mask-to-target brightness ratios

had an influence on the hemispheres x mask interaction (column 4, Table l)

 

1For individual subjects, the orthogonal comparison was made on
the arc sin transform of the observed prOportions. In this situation,
analysis based on single data points (angles) is made possible by the fact
that the error variance is known to be 821/n where n-number of observa-
tions making up each prOportion (100 in this case).

29
as well as on the over-all masking effect (column 3, Table l) was initially
disappointing in light of the original goal to find a relevant stimulus
parameter for controlling or increasing the magnitude of the effect. How-
ever, some work by Schiller (1969 - unknown to this writer at the time of
this research) and a l9—experiment report by Turvey (1973) established
rather conclusively that relative energy level between mask and target is
a totally irrelevant dimension for "central" (cortical) visual masking -
via the dichOptic procedure. The parameters that do influence cortical
masking will be discussed in later sections. 0n the other hand, they find
that "peripheral" (retina and neural pathways leading to the visual cortex)
visual masking is most dependent on the relative energy level between mask
and target - via the monOptic paradigm (the brighter the mask in relation
to the target, the stronger the effect). The failure to find group differ-
ences in this experiment supports, in view of the above findings, the
thesis that the observed masking was central (and therefore interhemis-
pheric in this case) rather than peripheral.

Briefly, two additional aspects of the between-groups comparison
deserve mention. For Groups I, III, and IV, mask intensity was held con-
stant while target intensity was decreased by 501 increments. One might
predict that the effect of halving target intensity might simply be that
of requiring a doubling of the target exposure time required to maintain
threshold responding levels. This statement is a parallel to Block's
law which states that the time and intensity of a light flash required to
produce a constant visual effect are reciprocally related. What was not
known, of course, was whether the same relationship would apply to the
detection of information from briefly exposed targets. If the law does

hold, then the total stimulus energy of the target would remain constant

30

as well as the mask-to-target ratio for Groups III and IV when compared
to I. Examination of the mean target exposure time required to maintain
threshold for Groups I, III, and IV (column 5, Table 1) reveals that for
Groups I and III the law holds approximately; but not so well for the
I and IV or the III and IV comparisons. A possible explanation for this
could be that if the light intensity of the bulbs was still rising over
the time intervals used, then a longer exposure would also be a more
intense one.1

For Groups I and II, target brightness was held constant while
mask brightness was decreased by 501 of the mask - target difference.
The numbers in column 3 (Table 1) show that over-all masking was somewhat
less for Group II (not significant) than for any other group which suggests,
according to the findings of Schiller (1968) and Turvey (1973), that some
small percentage of the variance attributable to masking may have been
"peripheral." The most plausible peripheral effect in this eXperiment
would be that due to stray illumination within the eyeball. (Lateral in-
hibition is discarded because of the large separation retinally between
mask and target.) Nevertheless, target exposure time to maintain threshold
remained unchanged, and more important, the differential effect between
hemispheres (column 4, Table 1) did not change. Obviously, it is difficult

to account for the differential masking effect with a peripheral hypothesis.

Error analysis 1
Central neurOphysiological or parallel "cognitive" mechanisms

that might account for "how" the masking effect is accomplished has not

 

lBrightness levels of the stimulus fields were measured with bulbs
constantly lit.

31
been a major (though admittedly important) theoretical issue of this
paper except for the following implicit assumptions which determined the
choice of the masking pattern itself. It was assumed that "cross-field"
visual masking could occur relatively early in the succession of central
processing stations, that is in the areas immediately succeeding the
primary (1?) visual projection area (18 and possibly 19). It follows,
if we may project Hubel and Wiesel's (1968) findings on the monkey to
man, that these areas are still involved in processing the pattern ele-
ments of a visual stimulusl. Consequently, the selected mask contained
pattern elements whose size and density, when superimposed on the target,
effectively "broke up" the raw stimulus elements of the target. Alterna-
tive masks would, when coded, be those which would produce confusability
of sounds, for example, or of meanings, for instance a matrix composed of
random alphabetic letters.

From the above, then, it was supposed that "masking" would some-
how degrade or decrease the amount of physical stimulus information avail-
able from the target pattern. If that is the case, it might be expected
that the "quality" of a subject's forced guess when an error is committed
might depend on the amount of physical stimulus information available to
him, and therefore on whether a particular trial was a mask or no-mask
condition. Specifically, it could be hypothesised that of the total

number of errors made, the percentage of errors that appear to be depend-

ent on physical stimulus information (report of letters with similar shape

 

1Not until the infero-temporal cortex has been reached (in monkey),
has it been found that receptive fields of individual neurons attain a
specificity for stimuli more complex than single lines, bars, and edges.
For example, the silhouette of a hand - more specifically, a monkey hand
(Gross et al., 1972).

32
but actually different from the target letters) might be higher under
the no-mask as compared to the mask conditions. 0n the other hand, if
masking affected primarily a distraction of attention, say, then errors
related to the similarity-in-shape dimension might distribute randomly
across mask conditions. It should, however, be emphasized that such an

analysis based on confusability of shapes per se does not permit dis-

 

crimination between central versus peripheral mechanisms as explanatory
constructs. It can be argued for example that a perceived decrease in
contrast between target and "ground" produced by stray illumination with-
in the eye could account for a decrease in information about target
features. Rather, it should support or reject the notion that some
process degraded the content of physical stimulus information contained
in the target.

For present purposes, then, all errors were divided into two
categories: (a) those in which there appeared to be a basis for physical
similarity between the letter guessed and the actual target, the stimulus-
dependent category; and (b) all others, the stimulus-independent category.
The letters classified as stimulus dependent (listed in Appendix B) were
selected without prior knowledge of experimental conditions into which
they fell1 and by the following criteria: similarities (judged by the
experimenter) based on addition or subtraction of a small number of frag-
ments, e.g. E and B, C and G and 0, F and P, etc.; similarities based on
"almost alike shapes," e.g. A and H, K and x, U and V; mirror image re-
versals, e.g. A and V, M and W, S and Z, J and L; and letter pairs which

upon examination of the response data showed a high frequency of

 

1The raw data sheets contained only the target letters, the re-
sponse letter for each target, and a "right" or "wrong" code for template

scoring.

33
confusability but could also be justified on the basis of some physical
similarity between these particular hand-drawn letters, e.g. H and M or W,
S and B, R and B, O and U, etc.

For each subject and each "within-subjects" condition the per-
centage of total errors falling into the stimulus-dependent category were
computed. These data averaged across subjects and groups are presented
in Table 3. To reiterate, the larger the percentage of stimulus-dependent
errors, the greater the stimulus (target) information is presumed to have
been received by the subject. Employing this new index as dependent vari-
able, an analysis of variance, identical to that reported earlier in this
section, was performed.

Table 3. Percentage of stimulus-dependent errors for left versus right
hemisphere and mask versus no-mask conditions averaged for Groups I-IV,

percentage decrement produced by the mask, average across hemisphere con-
ditions, and difference in percentage decrement.(Error analysis 1)

 

 

 

 

 

diff. in
T to T to average 2 decr.
left fl right R across R RB-LH
Groups I-IV no-mask 50.0 42.6 46.3
average mask 39.4 33.4 36.4 0.4
2 decr. 21.2 21.6 21.8

 

Note. T=target; H=hemisphere; R-right; L=left; 1 decr.-percentage decre-
ment produced by mask

The results showed that masking produced a 21.81 decrease (p4:0.001)1
in stimulus-dependent errors, and a left hemisphere superiority was ob-
tained (p4(0.0005). It was not expected that this index would prove
sensitive to left-right field differences. The within-subjects hemis-
pheres x mask conditions interaction term was not significant, but the

triple interaction was (p4<0.005). All other terms were not significant.

 

lAll significance levels reported for the error analyses are based
on the directly obtained F ratios.

34

The significant triple interaction showed that the hemispheres X masking
interaction differed for the varying mask-to-target brightness ratios.
Examination of this variation and also the over-all masking effect for
the different groups revealed no monotonic changes with increasing or
decreasing mask-target brightness ratios.1

The above finding that masking produces a significant decrement
in the percentage of stimulus-dependent errors supports the hypothesis
that the neuronal representation of the visual stimulus elements of the
target pattern has been degraded by the mask. Further support for the
contention that the present index reflects changes in stimulus degrada-
tion derives from the finding that the condition requiring the longer
callosal transfer route of target information to the left hemisphere

also produced a significantly lower percentage of stimulus-dependent

errors .

Error analysis 2

Even though the main effects variables were significant in error
analysis 1, the between-groups comparisons made no particular sense, and
consistency between subjects was quite poor. Although it is perhaps un-
reasonable to expect an alternate dependent variable to behave as nicely
as the original, it was thought, after reexamination of the original
criteria for defining stimulus-dependent errors, that the index might
be improved. It was noticed that many letter pairs which were similar
on the shape dimension afforded a great deal of confusability on the
auditory dimension as well, for example E and B, P and B, M and N, and
C and G. A second set of error categories was constructed (Appendix B)

in which auditorily confusable letter pairs were eliminated as well as

 

1Summary tables and the analysis of variance are available in
Appendix C.

35

mirror image reversals and any others which seemed to have a more tenuous
shape relationship. The final set of errors that could be classified as
stimulus-dependent guesses was reduced from the original 72 to 28 in
number.

The results, averaged across subjects and groups appear in
Table 4. An analysis of variance1 identical to the previous one was
performed. The findings were, in essence, identical to those of the
first error analysis. The masking effect was significant (p<:0.001),
Table 4. Percentage of stimulus~dependent errors for left versus right
hemisphere and mask versus no-mask conditions averaged for Groups I—IV,

percentage decrement produced by the mask, average across hemisphere con-
ditions, and difference in percentage decrement. (Error analysis 2)

 

 

 

 

 

diff. in
T to T to average 1 decr.
left H right B across fl (RH-LN)
Groups I-IV no-mask 29.8 26.2 28.0
average mask 23.7 18.3 21.0 9.7
2 decr. 20.5 30.2 25.4

 

Note. T=target; H-hemisphere; R-right; L=left; Z decr.- percentage decre—
ment produced by mask

the left hemisphere superiority was even more significant (p<:0.001 com-
pared to the previous 0.005), and the triple interaction remained signifi-
cant (p‘<0.05), though less so. The fourth column of Table 4 shows that
the difference in percent decrement, right minus left hemisphere, was in
the direction that would be predicted by the model and was as large in
magnitude, 9.7%, as that obtained with the original dependent variable,
probability correct. Nevertheless, the hemispheres X masking conditions
interaction was non-significant, and the between-subjects variation was

much larger for this index as compared with the original dependent variable.

 

1available in Appendix D

Recent writers have offered some interesting speculations about
the nature of the visual masking process. Turvey (1973) reviews two
kinds of masking processes originally elucidated by Kahneman (1968),
both of which have received wide support. The first view is that the
mask mixes with the "sensory" information of the target stimulus. "The
idea is that two stimuli which follow one another in rapid succession
are effectively simultaneous within a single frame of 'psychological'
time, analogous to a double exposure of a photographic plate" (Turvey,
1973). This is the integration hypothesis. The second, or interrup-
tion hypothesis, states that the (usually aftercoming) mask simply
terminates central processing of the target by occupying, so to speak,
the same processors. In essence, the time needed to completely process
the target has been cut short. While Turvey stipulates that these hypo-
theses need not be mutually exclusive, he views the integration process
as being primarily a function of "peripheral" mechanisms and inter-
ruption as a function of "central" processes. He supports these views
by the following argument: masking by interruption should depend prim-
arily on stimulus onset asynchrony (80A) or the onset-onset time rela-
tionship between target and mask, whereas integration should more likely
be controlled by relative stimulus energies of target and mask. He then
proceeded to demonstrate that SOA primarily affected central masking
(via the dichOptic paradigm) and that relative mask-target energy levels
affected only peripheral masking (via the monOptic procedure). Neither
variable was effective with the alternate procedure (within the range of
time and energy relationships employed).

Though relevant to the question of masking mechanism introduced

earlier, the basic model with which the above issues are concerned is

37
backward masking - a paradigm most useful for the study of information
processing. The model of this paper (as presented so far) differs
slightly in that it is concerned only with the limiting case of back-
ward masking, that is, simultaneous or overlapping coexistence of target
and mask representation in the central processors. The question of $2535-

ruption in central mechanisms presently gives way to one of integration,

 

and it is now obvious that the position taken for the present model (and

supported by the error analysis) is an integration hypothesis. The con-

 

cept of masking by integration requires that the iconic storage of the
target be contaminated at the input stage by sensory information from

the mask (Turvey, 1973) and, in the present case, it must be effected

 

centrally by separate neural input pathways. Can such a position be
rationalized on the basis of present neurOphysiological knowledge of the
visual system?

Recent work by Schiller (1968) provides a plausible explanation.
In this study, he attempted to find electrOphysiological correlates to
the visual masking process by measuring responses of single neurons in
the lateral geniculate nucleus and of binocularly driven cells in area
17 of the cat's visual cortex. After plotting the receptive fields of
individual neurons and studying their response to the target and then
target plus mask, he found that by making the contours of the pattern
mask more and more similar to those of the target, the cells failed to
respond differentially to the two stimuli in combination. In other words,
the target lost its separate "identity" in terms of the response charac-
teristics of the cell. His conclusion: "Pattern masking is brought
about primarily by a failure of units, whose receptive fields lie near

the contours of the figure, to respond differentially to target plus

38
mask and mask alone. It appears that this effect may take place either
cortically or at earlier levels in the visual pathway depending on the
mode of stimulus presentation. Under dichOptic conditions, the effect
occurs at the cortical level, while under monOptic or binocular condi-
tions it may take place earlier." (Schiller, 1968, p. 162.)

Behavioral evidence supporting a central (cortical) masking-by-
integration hypothesis comes from this experiment, of course, and the
studies by Schiller (1965) and Turvey (1973). As discussed earlier,
Schiller found that dichoptic masking, while successful with a pattern
stimulus, was ineffective when a bright flash of light was used as mask.
Further, Turvey (1973) was unable to obtain a dichOptic effect with a
random dot noise pattern. Only when the contours of the mask were re-
designed to simulate the contours of his target letters (a random array
of short straight lines with approximately the same thickness as the
target letter features), was he able to obtain a dichoptic effect.
These observations, then, that structural stimulus similarity between
mask and target are critical for dichOptic masking, support the concept

of "sensory" integration of stimulus elements at the cortical level.

Chapter 3: EXPERIMENT II

 

The results of Experiment I failed to satisfy one of the goals
set for this series of experiments, that was to find a physical stimulus

parameter that would control the magnitude of the differential cross-

 

field masking effect. While the differences in mask-target energy
levels failed to influence masking, this finding ironically (on the basis
of unknown, Schiller, 1968, or unavailable, Turvey, 1973, research) sup-
ported the hypothesis that the obtained effects were central rather than
peripheral. The purpose of Experiment II was to investigate the stimu-
lus onset asynchrony (SOA) variable, i. e. the onset-onset relationship
between target and mask, as being possibly relevant to the control of
the differential cross-field masking effect.1
The new model is basically simple in concept. Suppose, for pur-
poses of illustration, that the target and mask intensities to be used
are equal and that the callosal transfer time between hemispheres for
stimuli of the particular intensity is known to be 8 ms. If the onset
of a target in the left visual field precedes the onset of the mask in
the right half-field by 8 ms, then their arrival in the 15;; hemisphere
should be simultaneous (see Figure 9) due to the additional 8 ms required

for the target to transfer from the right to the left hemisphere. For

present purposes, the term stimulus arrival asynchrony (SAA) will refer

 

LThe finding of Turvey (1973) that central effects depended only
on SOA values (and not on relative energy levels) was not available at
the time this experiment was conducted.

39

40
to the relative arrival times between two stimuli for a given hemisphere.
In the preceding illustration SAA for the left hemisphere is zero. Pro-

ceeding now to the right hemisphere, the mask arrives 16 ms after the

target QD mask
l

O

t (1" to+ 8 ms

responding
hemisphere

 

 

 

 

SAA - 0 SAA ' 16 ms
(target preceding mask)

Figure 9. Expected stimulus arrival asynchrony of target and mask stimuli
at the two hemispheres for onset of a left-field target preceding onset of
a right-field mask by 8 ms with an assumed callosal transmission time of 8 ms.

Note. t - arbitrary onset time of first (target) stimulus; tl- onset time
of mask; SAA - stimulus arrival asynchrony

target (Figure 9) because (a) mask gppp£_was 8 ms later than that of the
target and (b) it takes the mask an additional 8 ms to cross from left to
right hemisphere. Consequently, the condition that obtains for the 15;;
hemisphere is one that simulates the limiting case (maximum effect) for
backward masking with a noise pattern (Kahneman, 1968), i.e. simultaneity
of target and mask. SAA for the right hemisphere now corresponds to the
backward masking paradigm which is known to produce a smaller effect than

when SAA is zero (Schiller, 1965; Kahneman, 1968; Turvey, 1973). Backward

41
masking with noise pattern is a monotonically decreasing function of SOA1
(Kahneman, 1968).

To follow the same reasoning as with the first model, we also
wish to directly assess the relative effect of the 16 ms SAA condition.
This is accomplished by reversing the target and mask positions of the
above example and maintaining the same target-mask onset-onset relation-
ships (Figure 10). NOw, SAA-0 for the right hemisphere, but more import-

ant, the mask follows the target by 16 ms in the left or responding

 

mask qp target
t1. t0+ 8 1118'

I

l

l

l

l

l

l

responding
hemisphere

 

 

 

 

SAA - 16 ms
(target preceding mask)

Figure 10. Expected stimulus arrival asynchrony of target and mask stimuli
at the two hemispheres for onset of a right-field target preceding onset

of a left-field mask by 8 ms with an assumed callosal transmission time of

8 ms.

Note. to- arbitrary onset time of first (target) stimulus; tl' onset time

of mask; SAA - stimulus arrival asynchrony

1In the dichoptic paradigm SOA is considered equal to SAA, as
defined here, for stimuli of equal intensities because the pathways are
of equivalent length.

42
hemisphere.1 The decrement in responding accuracy (no mask to mask)
should now be less than when SAA-0 for the responding hemisphere.
It follows that the conditions of Experiment 11 should be super-
ior to those of Experiment I for producing the desired differential mask-

ing effect. SOA for Experiment I was 50 ms, mask leading target. This

 

value is so much greater than any estimates of callosal crossover time
(6-10 ms, Efron, 1963; Jeeves & Dickson, 1970) that in effect mask should
overlap the target timewise in both hemispheres. That is, the mask
arrives before, is "on" during, and is present after the brief target
representation in both hemispheres. Hence, the only differential ad-
vantage would be that due to the predicted difference in signal-to-noise
ratios. It is prOposed that the conditions of Experiment II superimpose
a second advantage over that of Experiment I. With target in the left
and mask in the right fields, the low signal-to-noise ratio for the left
hemisphere is now combined with the simultaneity of arrival condition for
that hemisphere. This pair of conditions (for the left hemisphere) is
expected to produce little difference compared to Experiment I since
simultaneity is little different from overlap. (Forward masking produces
relatively small effects centrally, Kahneman, 1968; Turvey, 1973.) On
the other hand, the pigg signal-to-noise condition for the right hemis-
phere (or for left hemisphere with mask and target positions reversed) is
combined with the asynchrony in target-mask arrival time,and the combina-
tion should produce a smaller masking effect than for Experiment I. The
predicted result is a larger differential masking effect between hemis-

phere conditions.

1This hypothesis includes the implicit assumption that trans-
mission time of target information from right to left hemisphere is the
same as from left to right - and similarly for the mask.

43
The strategy for Experiment II was to select what appeared to
be the best of the four groups in Experiment I on the basis of differen-
tial masking effect, determine the apprOpriate SOA (to produce simultaneity
in arrival at the left hemisphere) for the exact mask-target brightness.

levels employed for that group, run a new group of four subjects with the

 

new SOA value, and compare the two groups. All conditions for the two
groups would be identical except for the different target-mask SOA
values. Group III from Experiment I was selected for comparison pur-
poses.1 All that remained was to determine the apprOpriate SOA value.

The report by Efron (1963a) provided the theory and methodology.

Determination of SOA value to produce left-hemisphere simultaneity

Efron (1963a) prOposes that for two stimuli traversing different
neural pathways to be judged as simultaneous in occurrence, they must
meet at some common point in the central nervous system. He reviews the
research of various investigators dating back to 1912 which shows that
(a) estimates of callosal transmission time deriving from various reaction
time experiments (stimulus to one hemisphere requiring response of either
the ipsilateral or contralateral hand) appear to be on the order of 2-6 ms
and (b) when bilaterally symmetrical points of the body are stimulated,
the left must precede the right by 2-6 ms in order to be judged as simul-
taneous for most peOple. He suggests therefore that the location for

the judgment of simultaneity must be the hemisphere that is dominant for

 

1Time and subject scheduling required that this selection be made
prior to the detailed statistical and error analyses. Although the four
groups did not differ statistically, it turned out that Group II would
have been a slightly better choice. Group II came closest to producing a
significant hemispheres X mask conditions interaction term when the groups
were analysed separately, and it produced a larger differential effect
under error analysis scrutiny than any other group. However, the overall
analysis suggests that for practical purposes, any of the four groups would
have sufficed.

44

speech (the left) since the left-right SOA corresponds to the time re-
quired for stimulus information to travel from the right to the left
hemisphere. He supported these contentions by showing in his own series
of experiments that for Eiggt-handers (presumed left hemisphere speech
dominant) a visual stimulus flashed to the left field had to precede a
stimulus flashed to the right field to be judged simultaneous. But for
left-handers (30-401 presumed right hemisphere speech dominant) the
right-field stimulus had to precede the left. In a companion paper
(Efron, 1963b) he determined that visual stimuli of lower intensity re-
quired a longer callosal transmission time than for higher intensity and
that for left-right field stimulus pairs of unegual intensity, the SOA
requiring a simultaneity judgment could be predicted by algebraic summa-
tion (or subtraction) of callosal transmission times for the respective
stimuli.

To determine the appropriate SOA value required for Experiment II,
Efron's (1963a,;nu 264-265) procedure was replicated on two graduate
students - but employing the brightness values used for Group III, Experi-
ment I. The average SOA value obtained would be used as a constant best
estimate value for all subjects of Experiment II. It should be noted
that for present purposes it is not necessary to obtain the callosal
transmission time for the less intense versus the more intense stimulus
and then calculate the 35355 SAA time for the right hemisphere. Rather,
it is only required to find the SOA value for the two stimuli to produce
simultaneity at the left hemisphere.

Method and results. Two right-handed male graduate students with
20-20 corrected vision and who were experienced at making delicate psycho-

physical judgments were run as subjects. Both were naive to the purpose

45
of the experiment. The apparatus was that used for Experiment I. The
stimuli consisted of two small back-lit circles, one in each visual half-
field. The diameter of each circle (prepared by cutting a hole in black
light-impervious paper) was identical to the height of the target letters;
and their location, equidistant from the fixation point, was identical to
that of the target letters which appeared (Experiment I) in the left and
right visual fields. The brightness of the left-field circle was always
0.20 ft.l. and of the right was 2.50 ft.l. (the brightness of target and
mask respectively of Group III, Experiment I). Each circle would be
flashed for a duration of 7 ms.

The procedure employed was (after Efron, 1963a, pp. 264-265) the
method of decending limits from alternative directions, i.e. left on
first, then right on first. The subject while viewing the fixation point
decided which circle came on first by responding "left-first," "right-
first," or "same." Each run started with, say, the left preceding the
right by 100 ms. For each succeeding trial, the interval was reduced by
5 ms steps until a simultaneity judgment was made. Following this, a new
run was started with the apposite stimulus appearing first. For any
interval setting, the subject could view the stimuli as many times as he
wished before making a judgment. After four practice runs, two from
each direction, each left-rightdescending;series was repeated ten times.
The starting points were random so as to vary the length of each series.

The average results for the two subjects were 11.5 ms and 9.25 ms
respectively - the left coming on before the right to obtain a judgment of
simultaneity. The between-subjects mean value of 10.38 ms was the SOA
value, target preceding mask, to be used for the four subjects of Experi-

ment II. The SAA value for the right hemisphere must be somewhat less

46
than 2x10.38 ms or 20.76 ms because the two stimuli are of unequal in-
tensity. A reasonable estimate based on Efron's work would place it

somewhere between 15 and 20.76 ms.

METHOD

Subjects

The subjects were four paid volunteers from the same two under-
graduate psychology classes from which the subjects in Experiment I were
drawn. All were males, right-handed, as determined by the Crovits and

Zener (1962) questionnaire, and naive to the purpose of the experiment.

Apparatus

Same as Experiment I.

Stimulus materials

Same as Experiment 1.

Procedure

The procedure was identical to that of Experiment I except for
the following: all subjects obtained the mask-target brightness condition
as for Group III of Experiment I, i.e. target-0.20 ft.l.; mask-2.50 ft.l.;
onset of the target field preceded onset of the mask field by 10.38 ms
for all four stimulus conditions. Consequently, for the "right hemisphere-
mask" condition (target to right hemisphere - mask to left), SAA for the
left (responding) hemisphere is estimated to be zero, and for the "left
hemisphere—mask" condition (target to left hemisphere - mask to right),

SAA for the left hemisphere is estimated to be between 15 and 20.76 ms.1

 

1As a limiting condition, it must be greater than 10.38 ms by the
amount of time it takes for the high intensity stimulus information to
cross the corpus callosum.

47

RESULTS AND DISCUSSION

The results for this experiment (Group V) as well as those for
Group III, Experiment I, are summarized in Table 5 and the "one between-
two within subjects" analysis of variance in Table 6. The nested between-

groups factor now represents the different SOA values.1

Table 5. Percentage of correct responses for left versus right hemis-
phere and mask versus no-mask conditions for Groups III and V, percentage
decrement produced by the mask, average across hemisphere conditions,
difference in percentage decrement, and mean target exposure time at
threshold.

diff. in mean T exp.

 

 

 

 

 

 

 

 

 

T to T to average 1 decr. time (ms) @
left H right H across H (RH-LN) threshold

Group III no-mask 67.5 57.0 62.3

SOA-50 ms mask 47.0 34.8 40.9 8.6 8.3

H before T Z decr. 30.4 39.0 34.7

Group V no-mask 53.3 57.3 55.3

SOA-10.38 ms mask 48.8 41.8 45.3 18.6 5.0

T before N. 1 decr. 8.5 27.1 17.8

Note. H-mask; T-target; H-hemisphere; R-right; L-left; SOA-stimulus onset
asynchrony; 1 decr.-percentage decrement produced by mask

Considering first the results of Group V alone (anova not tabled),
the difference in percent correct between hemispheres was not significant,
the masking effect was significant (p<<0.025), and the hemispheres x mask
conditions interaction was significant (p‘<0.01) in the direction predicted

by the model. It is noteworthy that this is the only one of the five groups

1Strictly speaking, the design requirements of this analysis were
not fully met, i.e. subjects were not randomly assigned to the nested
groups. However, since the subjects were drawn from the same classes
during the same term, it was not felt that there was any systematic factor
that could have influenced the content of the two groups. A better pro-
cedure, had it been practical, would have been to include four additional
subjects under replicated Group III conditions.

Table 6.

48

Analysis of variance summary

for Experiment II (Groups III and V)

Source

Total

Between subjects (8)
Groups4(A)

S/A

Within subjects
Hemispheres (B)

A;

SB/A

Mask conditions (C)
Ag

SC/A

BE

égg

SBC/A

 

°p < 0.05 (two-tailed)

gg
31
7
1
6

24

O‘HHO‘HHO‘HH

SS

4014.719
288.469
13.781
274.688
3726.250
331.531
195.031
602.188

1968.781
258.781

206.188
81.281
42.781
39.688

13.781
45.781

331.531
195.031
100.365

1968.781
258.781

34.365
81.281
42.781

6.615

based on percentage correct scores

2 5

0.301 (ns)

3.303 (ns)
1.943 (ns)

57.291 7.569***
7.5310

12.288 3.506**
6.468 2.543*

*p<0.025 (one-tailed); 1“:p< 0.01 (one-tailed); mp<0.0005 (one-tailed)

which produced a significant hemispheres X mask conditions interaction when

analyzed separately (on the basis of four subjects).

The most important comparison for the issues of Experiment II is

made in column 4, Table 5.

It is seen that the difference in percentage

decrement right minus left hemisphere increases from 8.6 to 18.6 with the

change in the SOA parameter.

To reiterate, the above index is a summary

number which expresses the magnitude and direction of the double interaction

(hemispheres x mask conditions).

The significance of the change in that

index betweenpgroups is therefore expressed by the triple interaction term

 

of Table 6 which is significant at the 0.025 level (one-tailed).

However, the manner in which the change in the double interaction

(i.e. the triple interaction) occurs is also important for present issues.

49
It was predicted that the primary change would be produced by a smaller
masking effect resulting from the SAA of 15 - 20.76 ms or backward mask-
ing condition at the left hemisphere for the target-to-left h. mask-to-
right h. condition. The third row for each group (Table 5) shows that
this was the case.

Perhaps a more lucid picture of this change, and the nature of
this one-tailed test of the triple interaction term can be given in the
following manner: for each subject we can find the difference between his
"no mask" and "mask" percent correct score for each hemisphere (the larger
the difference, the larger the masking effect). The average of those dif-
ferences, across subjects, appears in Table 7 (these values can also be
obtained by direct subtraction of the means in Table 5). A decrease from
Table 7. Average difference in percentage correct scores, between no-mask

and mask conditions, for left and right hemisphere versus two "stimulus-
onset-asynchrony" conditions

T to T to
122281281841
Group III
SOA-50 ms 20.5 22.3
N before T
Group V
SOA=10.38 ms 4.5 15.5
T before M

 

Note. H-mask; T-target; H-hemisphere; SOA-stimulus onset asynchrony

22.31 to 15.51 is obtained for the "right hemisphere" condition, while a
much larger decrease in masking effect, from 20.51 to 4.51, is observed
for the "left hemisphere" condition. The triple interaction has now been
reduced "in form" to a double interaction. The one-tailed test of the

orthogonal comparison of these four means, based on the a priori

50
prediction that the largest change in masking occurs in the "left hemis-
phere - mask" condition, is made in Table 8 (after Myers, 1967, p. 344).
Note that the F and t values in Table 8 are identical to those for the
triple interaction term in Table 6.
Table 8. One-tailed orthogonal comparison test of the interaction
between hemisphere and stimulus-onset-asynchrony conditions based on

the average difference in percent correct scores (no-mask minus mask
conditions)

____5°ur¢° ii £92. 115 I .t.
p(A) x q(B) 1 85.563 85.563 6.468 2.543’”
S x q(B)/A 6 79.375 13.229
*p40.025

 

Note. A-groups; B-hemispheres; s-subjects

From‘Table 6 a significant groups x masking (AC) condition inter-
action effect was also obtained. Columns 3 and 5 of Table 5 show that
this interaction is due to a decrease in the overall masking effect from
a 34.71 decrement to a 17.81 decrement with a change in the SOA parameter.
Also, the average target exposure time required to maintain threshold
decreased from 8.3 to 5.0 ms. Clearly, the targets were more detectable
'with target preceding mask by 10.38 ms. While most of this change is
reflected by the predicted decrease in masking for the left hemisphere
condition, a notable but smaller decrease from 22.31 to 15.51 (Table 7)
was obtained for the right hemisphere condition as wall. This result
could possibly be accounted for by the elimination of the forward mask-
ing eggpgnent of the total masking effect when the mask-target overlap
condition of Group III is changed to the SAA-O condition in Group V (for

the right hemisphere condition). While a central forward masking effect

51
does exist (Turvey, 1973), it is probably too small to totally account
for the above change.

What would seem to be a more likely hypothesis is that a possible
alteration in peripheral phenomena produced a small but £3221 decrease in
masking effect for pggh hemisphere conditions. As mentioned previously,
stray illumination within the eyeball is, for this paradigm, the most

likely source of peripheral interference between mask and target. (See

 

Bartley & Fry, 1934; Bartley, 1935; Boynton et al., 1954, for a discussion
of stray illumination phenomena.) This possibility seems especially likely
in the context of present experiments because of the semi-dark adapted
state of the eyes. However, the effect, whatever its magnitude, is ex-
pected to be equal with regard to the left or right hemisphere conditions
described here.1 If we now compare the stimulus conditions of Group III
with those of Group V, it can be seen that they differ greatly with
respect to possible stray illumination effects. For Group III, the on—
set of the 200 ms mask preceded that of the target by 50 ms so that for
both eyes, stray light was physically present during target exposure. For
Group V, however, target onset preceded mask onset by 10.38 ms and the
averege tagger exposure time was 5.0 T 1,3 ms. Consequently, stray illu—
mination from the mask could not have become physically present until
approximately 5 ms after offset of the target. In effect, whatever con-
sequence that could have resulted from stray light would have had to be

due to "backward masking" by stray illumination with an interstimulus

 

1The only laterality difference known to exist in the visual
system prior to the cerebral cortex is the difference in transmission
efficiency of the nasal versus the temporal hemi-retina of each eye. This
variable is, of course, held constant in the experiments reported here be-
cause both eyes are stimulated at the same time during any one stimulus
condition. A stimulus appearing in the left visual field, for example,
stimulates the temppral hemi-retina of the right eye and the nasal hemi-

retina of the left.

52
interval of 5 ms. In sum, it is prOposed that for Groups I to IV, a per-
centage of the total masking variance may have been due to stray illumina-
tion and that it made an equal contribution to both left and right hemis-
phere conditions. Thus stimulus conditions for Group V were such as to
considerably reduce the magnitude of a possible stray illumination effect
while producing an increase in the central differential effect by control-
ling the SAA intervals for the two hemispheres.

The general goals of the first two experiments were to establish
the tenability of the predicted left-right hemisphere differential mask-
ing effect and to identify the stimulus parameters that might be relevant
to increasing that effect. Both experiments provided statistical con-
firmation of the effect although its magnitude must, admittedly, be con-
sidered "small." Experiment I failed to show that mask-target brightness
relationships influenced either the overall or the differential masking
effect. The results of Experiment II established that the desired dif-
ferential effect could be increased by manipulation of the SOA parameter
and therefore the arrival-arrival intervals of target and mask informa-
tion at the two hemispheres. If it is assumed that the effects observed
in these experiments are "central" in origin, then the above findings are
in total agreement with Turvey's (1973) extensive analysis of peripheral
and central masking processes. Turvey found that masking via presumed
peripheral mechanisms was very insensitive to the SOA parameter while
for "central" masking, via the dichoptic paradigm, it was most crucial.
Experiment II is in agreement with those findings. On the other hand,
his finding that relative mask-target energy levels influenced only "peri-
pheral" but not "central" masking is in agreement with the results of

Experiment I .

53

While Experiments I and II were concerned with attempts to in-
crease the magnitude of the phenomenon, the remaining experiments deal
in a more direct manner with some theoretical issues about the neural
mechanisms involved. Experiment III tests the assumption about inter-
action between mirror-symmetric heteronymous areas of the visual half-
fields, and Experiment IV assumptions about single cell prOperties of
the neural substrate (and therefore the stimulus characteristics of the

mask), potentially responsible for the central masking effect.

Chapter 4: EXPERIMENT III

In the general introduction to this study it was prOposed that
the potential mechanism responsible for producing a masking effect bet-
‘ween the hemispheres was that which provided for continuity between the
left and right half-fields of vision and which also may be responsible
for stereoscOpic vision along the vertical meridian, i.e. the inter-
hemispheric connections through the spleneum of the callosum from areas
17 to 18 and 19. One of the features of this system as described by
BishOp and Henry (1971) and Blakemore (1970) is that the hemispheric
(stereosc0pic) interaction depends on stimulation of correspondipg
heteronymous retinal areas. The rationale for this requirement can
easily be seen by the following example. For a human fixating on some
object in his environment, objects which fall in front of or behind his
point of fixation but to his left or right side project sighs; to his
left or right hemisphere. Information from the disparate images of the
two eyes have a common meeting point (i.e. can fire a common cell) in
area 17 of the contralateral hemisphere. (See Bishop, et al., 1971,
for a discussion of how binocular interaction is coded by these units.)
On the other hand, objects which lie immediately in front of or behind
the fixation point along the vertical meridian stimulate corresponding
heteronymous areas of the two retinae and hence project to separate
hemispheres (refer to Figure 4). A stereOpsis effect resulting from
this particular kind of interocular disparity would, so the argument

goes, have to result from a combining of coded information from both

54

 

55

hemispheres via the corpus callosum. The earliest possible site for such
an interaction is, of course, the 18 side of the 17-18 border. As dis-

cussed in the general introduction, the behavioral support for this con-

 

tention was provided by Blakemore's (1970) study of a patient with a

sagittal section of his Optic chiasm, and neurOphysiological evidence

 

came from Berlucchi and Rizzolatti's (1968) demonstration that cells in

the apprOpriate part of area 18 could only be binocularly driven from

 

corresponding heteronymous areas (see Figure 5).
If this mechanism is responsible, at least in part, for the cross-
field visual masking phenomenon (which artificially simulates the above

stimulus conditions), it ought to be possible to show a difference in

 

masking effect produced by masks which are presented in correppondipg
versus non-corresponding heteronymous areas in relation to the target
stimulus. This was the goal of Experiment III.

In order to define corresponding versus non-corresponding visual
areas within the confines of the maximum stimulus field dimensions avail-
able in the tachistosc0pe used (2.130 wide x 2.900 high per visual half-
field after central occlusion for possible naso-temporal overlap), it was
necessary to substantially reduce the area of the noise pattern used to
about l/lOth the original size so that mask location could be varied.

The scheme for varying the location is depicted in Figure 11 for the right
field mask conditions. The geometric center of each mask is the same
distance from fixation as is the target on the opposite side. The mask

in position M-2 is considered the correppondipg heteronymous location

 

while those in positions M-1 and M-3 are defined as falling in non-cor-
responding heteronymous locations. The large mask (MeL), comprising the

entire visual half-field on either side, as well as the no-mask condi-

tions were retained for control purposes.

56

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 11. Corresponding and non-corresponding heteronymous mask loca-
tions for the right-field (left hemisphere) mask conditions.
Note. M-2 - corresponding; M-1 8 M-3 - non-corresponding

The predictions are that the small M-Z mask will produce the
same differential masking effect obtained in the earlier experiments
but that the "overall" effect (across left-right conditions) will be
smaller than that produced by the large mask due to a decrease in stray
illumination or possibly due to a decrease in total retinotOpic projec-
tion area represented in the visual cortex. Masks M-1 and M-3, while

providing equivalent extrafoveal stimulation, should produce a much

 

smaller or no effect compared to mask M-2 because they project to non-

corresponding heteronymous areas.

METHOD

Subjects

The subjects were four paid graduate students in non-perception
areas of experimental psychology at Michigan State University. All were
right-handed males with 20-20 or better corrected vision in each eye and

naive to the purpose of the experiment.

57

Apparatus

Same as for previous experiments.

Stimulus materials

The set of 100 target letters used in the previous experiments
were retained unaltered as well as the no-mask and large mask stimuli.
The small masks were prepared by photographing a cropped version of the
original noise pattern so that the entire stimulus slide was "black"
except for the mask appearing in its appropriate location. Six sets of
the new mask stimuli were prepared, three with masks appearing in the
M-1, M—2, or M—3 positions in the right visual field (see Figure 11),
and three for the same positions in the left field. The dimensions of
each small mask were 0.850 in width by 0.74° in height - a little more
than twice the height and width dimensions of the target letters and
approximately l/lOth the visible area of the large (M—L) masks. The
geometric center of these masks fell on an imaginary circle with a
radius of 1.70 measured from the fixation point (the same distance as
the target centers from the fixation point). The angular elevation of
the M-1 masks (left or right field) was 38.2° from the horizontal and
depression of the M-3 masks from horizontal was of the same magnitude.

The masks M-1 and M-3 together comprise the stimuli for the
non-corresponding mask condition. Consequently, there were half as many
Mel or M-3 slides (25 each for each field) as there were M-2 slides
(50 for each field). As a result, the total number of stimuli for each
condition (non-corresponding, corresponding, no mask, and large mask)

was the same.

58

Procedure

The general procedure was the same as that for the preceding
experiments except as noted below. The mask-target brightness ratios
and the controlled SAA conditions were the same as for Group V (that
group having produced the best results) except that the apprOpriate
SOA value was independently determined for each subject by the psycho-
physical procedures described in Experiment II.

The four within-subjects conditions defining the previous ex-
periments have here been expanded to eight. They included the no-mask,
large mask, non-corresponding small mask, and corresponding small mask
conditions for both the left and right visual half-fields. The random-
izing and counterbalancing procedures for presentation of target and
mask stimuli were the same as described earlier. Any one of the (now 4)
lOO-slide rototrays (for masking) contained either 12 or 13 stimuli for
each of the eight conditions. Administration of Egg 100 trial sets gave
the subject 25 trials for each of the eight conditions with each of the
25 target letters appearing once per condition. Excluding Day 1, which
was used for the psychophysical determination of a subject's SOA value,
a subject received 200 trials per day on each of four separate days
(usually within one calendar week). Consequently, the four different
sets of 100 masking stimuli were administered twice but in reverse order.
In sum, each unfortunate subject received a total of 100 trials for each

of the eight conditions.

59

RESULTS AND DISCUSSION

The percentage of correct reaponses averaged across subjects for
each of the eight conditions appear in the first two columns of Table 9.
The averages across hemispheres for each of the masking conditions are
tabulated in column 3. Results of the 2 x 4 within-subjects analysis of
variance are summarized in Table 10.
Table 9. Percentage of correct responses for left and right hemisphere

conditions as a function of the no-mask, large mask, small corresponding,
and small non-corresponding mask conditions of Experiment III (Group VI)

 

 

 

 

T to T to average
left H right H across H
no mask 49.5 53.0 51.3
large mask 38.8 37.0 37.9
(M-L)
small mask 48.3 50.3 49.3
(M-Z)
small mask 47.5 54.8 51.1
(M-1-3)

 

Note. T=target; H=hemisphere; M-2-corresponding; M-1 and M-3-non-corre-
sponding

Table 10. Analysis of variance summary based on percentage correct
scores for Experiment III

 

 

Source pg §§ MS E
Total 31 1377.500
Hemispheres (A), 1 60.500 60.500 3.924 (ns)
Mask conditions (B) 3 982.750 327.580 26.568 (p<E0.001)
Subjects (8) 3 40.750
A}; 3 83.250 27.750 4.711 (p40.05)
AS 3 46.250 15.420
BS 9 111.000 12.330
ABS 9 53.000 5.890

60

A difference between masking conditions (averaged across hemis-
pheres) was obtained (p410.001), but examination of column 3, Table 9,
reveals that the effect was attributable entirely to the effect of the
original "large" mask. Percentage correct for the no-mask, small cor-
responding, and small non-corresponding mask conditions was 51.3%, 49.32,
and 51.12 respectively, while for the "large" mask it was 37.91. Table 11
shows the result of all possible multiple comparisons between masking con-
ditions (by the method of Scheffé). A11 paired comparisons of each con-

dition with the large mask condition proved significant, while compari-

 

sons of the other conditions with each other were not. Clearly, in terms
of "overall" masking effect, the small corresponding and non-correspond-
ing mask conditions did not differ, and in fact neither of the "small"
mask conditions produced any effect at all.

Table 11. Statistical tests of the paired comparisons of all possible

combinations of masking conditions (averaged across hemispheres) for
Experiment III by the method of Scheffé

no mask --

M-L p 40.05 -_

H-Z ns p40.05 --

M-l-3 ns p40.05 ns --

 

no mask M-L M-2 M-l-3

Note. M-L-large mask; M-2-small corresponding mask; M-l-3=sma11 non-
corresponding masks
The results in terms of the differential masking effect (hemis-
pheres X mask conditions interaction) were similar. Although the over-
all hemisphereij mask conditions interaction term was significant

(p'<0.05), examination of any two rows (within the rectangle) in

61
Table 9 or of the breakdown in Table 12 showing all possible interactions,
reveals that none of the interactions of interest were significant. The
no-mask versus large mask interaction was in the direction predicted by
the model but non-significant, and the non-significant no-mask versus
the M92 mask interaction was in the direction predicted by the model,
but so small in magnitude (2.7% difference in percent decrement) as to
be totally inconsequential. The only interaction proving significant,
the large versus small non-corresponding mask conditions, is both unex-
plainable and unimportant in terms of present considerations.1 In sum,
the hypothesis of "no difference" between corresponding and non-corre-
sponding heteronymous mask locations (as here defined) cannot be rejected,
and some aspect of the pigg difference between these particular "large"
and "small" masks seems to have been crucial for obtaining the cross-
field masking effect.
Table 12. Statistical tests of the two-way interaction effects between

left-right hemisphere conditions and all possible pairs of mask condi-
tions for Experiment III by the method of Scheffé

no mask --

M-L ns --

M-2 ns ns --

M-1-3 ns p40.05 ns --

 

no mask M-L M-2 M-l-3

 

Note. M4L-1arge mask; M-2-small corresponding mask; M-l-3-sma11 non-
corresponding masks

 

1A comparison of rows 1,2, and 4 of Table 9 shows that the M—1-3
mask condition deviated "away from" the large mask condition in the same
direction as did the no-mask condition - only slightly more so for some
reason. The fact that the small M-l-3 masks did not differ from the pp-
mask condition, either in term of the overall or the differential inter-
action effect (as neither did the large mask condition), renders the ob-
tained "significant" interaction to a position of little theoretical
importance.

 

62

Before discussing the implications of these findings, it is
important to note the result of some methodological differences between
this and the previous experiments. The effects obtained for the control
(no mask and large mask) conditions were considerably different from
those obtained for Group V, Experiment II, even though the stimulus para-
meters and mask sizes were the same. Principally, the magnitude of the
differential masking effect, right minus left hemisphere, was much
smaller at 8.61 than the 18.6% value obtained for Group V. The only
controlled difference between the two situations is that for Experiment III
the no-mask and large mask conditions were "imbedded" within the two addi-
tional within-subjects conditions. The writer speculates that the above
result is due to the effect of more extended "practice" made possible by
Experiment III. The comparison between any subset of four conditions,
based on x number of trials per condition, involves the administration
of twice the total number of trials under the conditions of Experiment III
as compared with Experiment 11. The first point is that, as mentioned
earlier, left—right field differences obtained for many repeated ex-
posures of single alphabetic letters under the no-mask condition are not
especially reliable.l This was observed during pilot work for those
studies which involved administration of 400 - 1,000 trials to some sub-
jects. More significant support, however, stems from Schiller's (1965)
research which systematically investigated the effects of practice on
masking. He found that the effect of central masking by noise pattern

(via the dichoptic procedure) considerably decreased with practice over

 

1The reasons for this may involve a possible shift in the mode
of processing these highly overlearned pattern stimuli under conditions
of extended practice. Such shifts are known to occur (Grill, 1971) under
practice and they may relate to hemispheric differences in the pre-verbal
pattern recognition stages of processing.

63

trials and days. On the other hand, masking by the monOptic procedure,
which favors interaction by peripheral mechanism, was most insensitive
to practice effects. It would seem that as we go from peripheral to
more central neural mechanisms, there is an increased "adaptability" to
adverse stimulus conditions.

Having employed the control conditions as "yardstick," it would
seem, then, that Experiment III was not optimally designed to measure
small changes. Spreading a given number of observations too thinly
over so many within-subjects conditions seems to have favored a greater
influence by unwanted (for present purposes) practice effects. Employ-
ment of more complex stimuli (e.g. words) would, of course, help allevi-
ate this problem since the larger field differences obtained (Kimura,
1973) can be reliably measured with a smaller number of observations.

In light of the above, it was decided to substantiate or re-
check the unexpected finding that the small mask (M-2), in the correspond-
ing heteronymous location, has no effect by replicating part of Experi-
ment III under conditions identical to the preceding experiments. Three
additional subjects were run under the same conditions as Experiment V
(four within-subjects conditions at 100 observations per condition,
mask - 2.40 ft.l., target - 0.20 ft.l., SOA - a constant 10.38 ms for
all subjects) except for replacement of the large mask by the small M-2
mask. The results for each subject are shown in Table 13. Subject J.B.
showed no masking effect at all. The remaining two subjects produced a
miniscule effect but in 2253 cases the effect obtained resulted from
their performance on the first 100 trials only. §g_effect was produced
on the last 300 trials by any subject. Though a small practice effect
is in evidence here, the inescapable conclusion is that the "small area"

Mr2 pattern is an ineffective cross-field masking stimulus.

64

Table 13. Percent correct responses for left-right and mask - no-mask
conditions for three individual subjects receiving only the small "corre-
sponding" mask stimulus, percentage decrement produced by the mask,
average across hemisphere conditions, difference in percent decrement,
and mean target exposure time at threshold

diff. in mean T exp.

 

 

 

 

 

 

 

 

 

 

 

 

 

T to T to average 1 decr. time (ms) @
left H right H across H (RH-LH) threshold
Subject no-mask 55.0 40.0 47.5
J.B. mask (M-2) 56.0 41.0 48.5 -0.007 5.1
I decr. —0.018 -0.025 —0.022
Subject no-mask 42.0 53.0 47.5
J.E. mask (M-2) 38.0 48.0 43.0 -0.1 5.7
I decr. 9.5 9 4 9.5
Subject no-mask 50.0 54.0 52.0
J.B. mask (M—2) 48.0 46.0 47.0 10.8 3.5
1 decr. 4.0 14.8 9.4

 

Note. All other stimulus conditions the same as for Group V.

M-2-sma11 mask in corresponding heteronymous location; T-target;

H-hemisphere; R'right; L-left; Z decr.-percentage decrement pro-

duced by mask

Failure to find a difference between corresponding versus non—
corresponding heteronymous mask locations for these particular stimuli
has two possible explanations. Either there 2;: no "true" differences
or the hypothesis was not adequately tested by Experiment III due to the
fact that mask M-2 was ineffective. The test of the above hypothesis was,
of course, predicated on the assumption that simply reducing the area of
the "large" corresponding heteronymous mask would not have the effect of
erasing the phenomenon. Obviously there must be a phenomenon before
differences can be measured. Consequently, the position taken here is
that the corresponding versus non-corresponding issue has not been adequate-
ly tested.

The question of central importance now becomes: "Why did reduction

in area eliminate a measurable effect?" Reduction in total eneggy level

 

65

per se (i.e. reduction of the total light flux - not flux density) by re-
ducing the area of the large mask does not seem a plausible explanation
for elimination of the effect. The findings of Experiments I and II
which are in agreement with those of Turvey (1973) simply do not support
such a contention. Experiment I showed that varying the energy level by
controlling brightness had negligible effects on the phenomenon; and
Experiment II showed that, although the "overall" effect could be lessened
by a set of timing conditions which substantially reduced the potential
interplay by peripheral mechanisms, the 23335 phenomenon remained intact
and even improved in terms of the "differential" criterion. In con-
sequence, it seems that alteration in total area of retinotopic projec-
tion (of the mask) to the visual cortex would be a more likely accounting.
Upon reconsidering the work of Berlucchi and Rizzolatti (1968)
and Hubel and Wiesel (1965), it was realized that a serious error may
have been committed by trying to imitate one feature of Schiller's
dichOptic masking experiments. The size of his masking pattern was about
1% times the size of his single target letters, and so it was assumed that
for Experiment III masking patterns a little more than Egigg the length
and width dimensions of the targets would be adequate. However, there is
a very important difference between the dichoptic and cross-field visual
masking paradigms: the probable masking sites within the visual cortex
and the response characteristics of the cells populating those sites.
It will be remembered that for the dichoptic procedure corresponding
homonymous areas of the two retinae are stimulated and the first central
location where the interocular information can meet at a common point is
area 17 (where single cells can be binocularly activated or inhibited).

Area 17 is most densely pepulated with "simple" cells which have the

66

characteristic of being, for all practical purposes, "point specific;"
that is the width of the receptive field sizes is on the order of 8° or
less (rhesus monkey, Hubel & Wiesel, 1968). This means that if the edge,
bar, or line is moved by more than, say, 8° in a direction orthogonal to
the preferred stimulus orientation, the cell ceases to be activated. In
light of this, it is easy to see why a mask pattern only 18 times the
size of the target is a more than adequate interference stimulus in area
17 (in terms of Optimally stimulating the relevant projection area).

For the "cross-field" paradigm, on the other hand, area 18 is
the first possible meeting site for stimulus information deriving from
corresponding heteronymous areas (see Figure 1). Area 18 is most densely
populated with "complex" cells, and their receptive fields are many times
larger. In Berlucchi and Rizzolatti's (1968) split-chiasm cat experiment
‘211 cells in area 18 that could be binocularly driven through the corpus
callosum (from corresponding heteronymous retinal areas) (a) were "complex,"
(b) had receptive fields which always extended to the vertical meridian,
and (c) had receptive field sizes which ranggd from 1 x 2° to 9 x 9°.
The prOperties of a complex cell are briefly as follows: a line of proper
orientation with length x (where x is smaller than the length or width
dimension of the receptive field) will fire the cell when moved anywhere
within the receptive field of that cell.1 Furthermore, Hubel and Wiesel
(1965) observe that increasing the length of the line up to the boundaries

of the receptive field continuously increases the response of the cell.

 

Increasing the length beyond the receptive field boundary has no addi-

tional effect.

 

1Hubel and Wiesel speculate that this characteristic of complex
cells may be a means of providing a stable image representation from eyes
that are continuously in motion.

67

Based on these facts, it is quite plausible that a mask pattern
whose area is considerably smaller than the receptive fields of the cells
occupying the presumed masking site will stimulate those cells 1513 than
would a pattern of larger area. Receptive field sizes of complex cells
in man can, at present, only be estimated from data on the rhesus monkey.
Hubel and Wiesel (1968) find those sizes to be about 1/4th the size (in
linear dimensions) of those of the cat which would make them on the order
of about 2 1/40 x 2 1/40 for the larger ones. The dimensions of the small
mask used in Experiment III were 0.850 x 0.740 while dimensions of the
larger one were (quite fortuitously) 2.1250 x 2.90. Also, the small
masks did not extend as close to the vertical meridian as did the large
ones - another receptive field prOperty of the binocularly driven complex
cells found by Berlucchi and Rizzolatti (1968). The above contentions
could at least be supported by conducting an experiment which systematic-
ally varied mask area while holding total brightness level constant. If
the above is tenable, then masking effect should be found to be an in-
creasing monotonic function of increasing mask area up to some optimal
point and then asymotote for larger sizes. The "near zero" point on the
curve has obviously been demonstrated by Experiment III.

Before concluding, we must return briefly to the corresponding
versus non-corresponding areas issue. If in fact the receptive field
sizes of the binocularly driven complex cells in area 18 are on the
order of 2° x 2° or greater, then the masks M-1-2-3 each stimulated a part
of the same corresponding heteronymous area, neurologically speaking.

In other words, the "geometric" definition of corresponding versus non-
corresponding for these particular masks and their relative locations may

have been irrelevant to the neural processors. If it should turn out

68
that the minimal-sized mask pattern required for producing an effect
should extend close to the vertical meridian, then it will be most diffi-
cult to directly test the corresponding versus non-corresponding issue.
The reason is that it would be impossible to define, within the two-
dimensional space of a visual half-field, non—corresponding areas that
‘ would provide equal retinal stimulation with respect to their distance
from the fovea.

The preceding line of reasoning regarding the role of receptive
field sizes of complex cells in area 18 leads to predictions about what
might have happened had certain conditions of EXperiment III been set up
differently. For example, in Experiment III the "white" field area sur-
rounding the target might have been decreased to match the area of the
"small" mask. Alternatively, a "white" background surrounding the "small"
mask and equal in size to that of the target field could have been used.
By the arguments developed above, no masking should occur, because the

critical feature in both situations should be the pgpportion of the recep-

 

tive field area (for cells in area 18) stimulated by pattern features, not
the relative background dimensions of target and mask stimuli. As these
combinations have not been tried, future research will have to resolve the
validity of these predictions.

Finally, how do these notions stand with the theory of binocular
stereopsis for viewing objects along the vertical meridian? What of the
concept of corresponding heteronymous ppipgg when we are dealing with cells
whose receptive field dimensions are sometimes as large as 20 x 2°? Several
comments on this line are in order. First of all, it should be recognized
that a 2° x 2° area is a very small portion of the total visual field.
Secondly, real life objects that may fall along the vertical meridian de-

fined by the fixation point are not themselves dimensionless points but

69

occupy some space across the meridian. Thirdly, such objects may deviate
slightly to the left or right of vertical as a result of being in motion,
and fourth, the vertical meridian itself is constantly being re-defined
by involuntary saccadic eye-movements. To derive order from such chaos,
the requirement of receptive fields which permit cross-field interaction
anywhere within a 2° x 2° space, does not seem extravagant.

The final experiment provides a direct test of the stray illumi-
nation hypothesis and of yet another assumption about the prOperties of

the cells involved in the masking interaction.

Chapter 5: EXPERIMENT IV

Aside from establishing the existence of the cross-field visual
masking phenomenon and determining the stimulus parameters which influ-
ence its magnitude, a central concern of this paper has been to establish
the tenability that the observed "effect" is a result of what the model
says it is instead of the result of something else. Specifically, the
ggdgl predicts a differential masking effect that results from a "central"
interaction between hemispheres on the feature content of visual stimuli.
The "something else" says that the effect is somehow due to peripheral
mechanisms (or from some "higher order" process such as distraction of
attention). The peripheral mechanism of greatest concern has been the
known existence of stray illumination resulting from the flash of a bright
masking pattern. While it has been conceded that stray illumination may
play a role in contributing to the "overall" masking effect (which was
actually supported by Experiment II), it has been argued that it cannot
account for the differential effect. Although various kinds of evidence
from Experiments I and II strongly supported the Operation of "central"
as Opposed to "peripheral" mechanisms, the peripheral (or stray light)
hypothesis has not been tested directly. It may be that the peripheral
mechanism £2219 produce the differential effect in some manner unanticipated
by the experimenter. The purpose of Experiment IV was to test the stray
illumination hypothesis by replacing the original mask with a stimulus

(a) which would provide an equal amount of stray illumination, (b) whose

70

7l

prOperties are known to produce a peripheral masking effect, and (c) which
is purported to have little or no effect on the contour detectors which
comprise the neuronal make-up of the visual cortex - that is, diffuse light.
In addition, the results of such a test should help confirm or reject the
idea that it is the feature detecting system in areas l7, l8, and 19 that
is responsible for the differential masking effect predicted by the model.1

That the neuronal information processors of the visual cortex are
relatively insensitive to stimulation by diffuse light derives support from
several sources. As mentioned in the general introduction to this report,
Hubel and Wiesel (1965) find that simple, complex, and hypercomplex cells
which constitute the neuronal make-up of areas 17, 18, and 19 respond
either very weakly or not at all when stimulated by a field of diffuse
light. One of the reasons for this, as the theory goes, is that a field
of diffuse light simultaneously activates both the excitatory and inhibitory
areas of the receptive fields of "dark" and "light" spot detectors which
presumably converge their information onto the more complex edge and line
detectors. The end result apparently is cancellation of any appreciable
neural response.

Behavioral evidence, as also mentioned earlier, is provided by the
visual masking experiments of Schiller (1965). Diffuse light was found to
produce a considerable peripheral masking effect, as when target and a light

flash are exposed to overlapping areas of the same eye (s). However, and

 

1While it is, of course, not known how this system works to effect
the "experience" of pattern recognition, it is indisputable that these
single cells detect fragments of visual pattern stimuli. Also, it should
be noted that "a few" such cells have been recorded in the visual cortex
of man, and they were found to have the simple, complex, and hypercomplex
prOperties originally described by Hubel and Wiesel (Marg et al., 1968).

72
in sharp contrast to the effect of a pattern, it was found to produce no
effect dichOptically; that is, when target and mask information first
meet in the "primary" cortical projection area (17). Others (Battersby
& Wagman, 1962, and Boynton, 1961) find a "slight" dichOptic effect when
target stimuli are near threshold and target and mask borders are in
"relatively close" proximity.

Finally, Spehlman (1965) has demonstrated differences in visual
cortex responses to diffuse light versus pattern stimuli by the gross
evoked potential techniques (EEG—type recordings from surface electrodes
on the scalp). Cortical responses (occipital lobe of humans) differed
not only in waveform but in magnitude to the two kinds of stimuli. When
amount of contour information was varied, he found that magnitude of the
evoked responses was directly prOportional to the "density of contrast
borders" - diffuse light being the low point on the scale.

The above paragraphs should not be taken to imply that the response
of the visual cortex to diffuse light stimulation is zero; far from it.
A case in point is the "photic driving response" (obtained from scalp
electrodes over the occipital region) to strobe-light flashes used in tra-
ditional diagnostic electroencephalography. Unfortunately, it is not known
what electrical events contribute to such changes in scalp potential.
Rather, the point is that this area of the brain appears, in humans, to
be restricted to the initial processing and perception of pattern stimuli
(see for example Stone & Freeman, 1973). In fact, the "experience" of
diffuse light apparently does not depgnd on the visual cortex at all
(cortical ablation and brain damage studies, ibid., 1973).

In a manner similar to Experiment II one of the previous groups

was selected for purposes of comparison with the results of the present

73
experiment. In this instance, Group V was chosen because the conditions
for that group produced the largest differential masking effect when the
noise pattern was used as masking stimulus. Consequently, the procedures
and stimulus parameters employed for Experiment IV were identical in every
aspect to those of Group V (Experiment 11) with the single exception that
the noise pattern was replaced with an equally bright diffuse light flash
on masking trials. The predictions were that 229: "masking effect" would
be produced - hypothetically resulting from stray illumination and
possibly cortical mechanisms other than areas 17, 18, and 19; the overall
masking effect would be smaller than that produced by the pattern; and

the crucial differential masking effect would disappear.

 

METHOD

Subjects

The subjects were four paid graduate students in various non-
perception areas of psychology at Michigan State University. All were
right-handed males with 20-20 or better corrected vision in each eye and

naive to the purpose of the experiment.

Appgratus

Same as for all preceding experiments.

Stimulus materials

 

A new set of "masking" slides were prepared which were "black"
on one half and transparent on the side which previously contained the
noise pattern. Each of the pattern mask slides that were used for
Group V (and also Groups I-IV) were replaced on a one for one basis with

one of the new slides. Consequently, on a "masking" trial for the present

74
experiment, the subject saw a contourless flash of light (except for the
outside rectangular boundaries) in the same location where he had previous-
ly seen the pattern mask. The spatial dimensions defining the boundaries
of the masking field were, of course, the same. All other stimulus
materials and their order of arrangement etc. remained unchanged from

what was described for Experiments I and 11.

Procedure

The physical features of the previous pattern mask were such that
501 of the total mask area was "black" and 50% was "white." In order to
produce a patternless light flash of equivalent overall brightness, it

1’2 This was

was necessary to halve the luminance of the masking field.
most easily accomplished by using the same intensity setting for the
masking channel that was employed for Group V (i.e. maximum) and insert-
ing a Kodak 0.30 neutral density filter in the masking field. This filter
has 50% light transmission characteristics.

Except for the modification just described all other experimental

procedures and stimulus parameters, including the use of the constant

SOA value of 10.38 ms for all subjects, remained the same as for Group V.

 

1While it is recognized that the experience of "brightness" is
not an absolutely linear function of stimulus intensity, it was felt that
the above procedure was a close enough approximation for present purposes.
2Halving the luminance or light intensity halves the "flux density"
which is equivalent to the average effect (over the entire field) of the
501 black pattern stimulus.

 

RESULTS AND DISCUSSION

75

The results of this experiment (Group VII) and those of Group V,

Experiment II, are presented in Table 14.

A comparison "one between -

two within subjects" analysis of variance1 is reported in Table 15.

Table 14. Percentage of correct responses for left-right hemisphere and
no-mask - mask conditions for pattern versus light flash masking stimuli,
percentage decrement produced by masking, average across hemisphere con-
ditions, difference in percentage decrement, and mean target exposure time

at threshold

 

 

diff. in mean T exp.

 

 

 

 

 

 

 

T to T to

left H right H
Group V no-mask 53.3 57.3
pattern mask mask 48.8 41.8
1 decr. 8.5 27.1
Group VII no—mask 50.0 57.2
light flash mask 33.8 44.8
1 decr. 32.4 21.7

*26.02 decrease from 5.0 ms

 

 

 

average 1 decr. time (ms) @
across H (RH-LH) threshold
55.3
45.3 18.6 5.0
17.8
53.6 *
39.3 -10.7 3.7
27.1

Note. Brightness and SOA conditions identical for Groups V and VII.
T-target; H-hemisphere; R-right; L=left; Z decr.-percentage decrement

produced by masking

The difference in percent decrement (right minus left hemisphere)

changed from 18.62 for the pattern mask condition to ~10.7Z for the bright

flash condition! This effect was significant (p<:0.05, two-tailed) as ex-

pressed by the triple interaction term.

Not only did the hemispheres X mask

conditions interaction change in a direction toward equality (i.e. no inter-

action), as predicted, but the diffuse light condition produced what must

 

1As was the case for Experiment II, subjects were not randomly
assigned to the two groups. Additionally, the composition of Group VII
differed slightly in both age and academic status in comparison to Group V.
However, the psychology graduate students that were run in the pilot
studies showed the same phenomenon with the pattern mask as did the under-

graduates in Groups I through V.

 

Table 15.

76

for Groups V and VII of Experiment IV

Source

Total

Between subjects (S)

Groups (A)

S/A

Within subjects
Hemippheres (B)
Ag’

SB/A

Mask conditions (C)
A§_

SC/A

BE

épg.

SBC/A

N
90‘

C‘Ht—‘O‘HHO‘HH

SS

2787.219
314.969
116.281
198.688

2472.250
116.281

225.781
450.688
1188.281
38.281
241.188
26.281
108.781
76.688

!§

116.281
33.115

116.281
225.781
75.115
1188.281
38.281
40.198
26.281
108.781
12.781

29.
.952

I'll

.512

.548
.006

561

2.056

.511

Analysis of variance summary based on percentage correct scores

(ns)

(ns)

(ns)

(p (O. 005)
(ns)

(ns)
(p (0.05)

be considered a sizeable interaction in the Opposite or negative direction.

(When the results of Group VII were analyzed separately, this interaction

in the Opposite direction was not statistically significant.) Three of

the four subjects in this group produced the above effect quite markedly,

'while the fourth showed a small interaction trend in the direction produced

by the pattern mask.

This "negative" interaction trend was first observed when the

results of the first daily session of 200 trials were scored and recorded

for each of the first two subjects.

This striking trend immediately

aroused suspicion that some sporadic alteration of the equipment may have

occurred which could affect the left-right luminance conditions in one of

the stimulus channels - especially the masking channel.

This was always

a possibility since the left or right member of each vertically mounted

 

77

pair of bulbs (though carefully matched at the onset) could deteriorate
at a different rate. This possibility was controlled for by reversing,
left for right and right for left, the fluorescent bulbs and the vacuum
tubes which independently drive each bulb for all stimulus channels.
Consequently, any idiosyncrasy that could have affected left-right field
conditions was now reversed. The 1253 200 trials were then administered
to the first two subjects with the altered apparatus. Their performance
did not change.

The other prediction made regarding the relative effects of the
pattern versus diffuse light stimuli was that the flash would produce a
smaller overall masking effect than the pattern. As previously, there are
two indices which reflect this change: the overall percentage decrement
(averaged across hemisphere conditions) obtained at threshold, and the mean
target exposure time required to maintain threshold across all conditions.
The first index (column 3, Table 14) showed an increase in average percent
decrement from 17.81 for the pattern mask to 27.1% for the light flash.
This change was not statistically significant (between groups factor,
Table 15), and it is seen (column 1, Table 14) to result from the increase
in percent decrement from 8.51 to 32.41 for the left hemisphere condition
(which also accounts for the shift in direction of the interaction). On
the other hand, the average target exposure time to maintain threshold
decreased from 5.0 ms for the pattern condition to 3.7 ms for diffuse
light - a 26.0% decrease. (The latter mean includes a value of 5.1 ms
obtained for the subject producing a position interaction.) This is a
substantial decrease because experience proved that a change by as few as
two 5% increments in target exposure time in either direction usually took

the subject completely out of threshold responding level - to either 100%

IVI
'— a.

a .56: fl).‘l.'ni'vr¢ols-e vine-Ir} ape-1a
‘1

 

.11) I ma“

78
or 0% correct. Stated in different terms, if Group VII had been run at
a fixed target exposure time of 5.0 ms, no masking effect would have been
observed, and average percent correct would have been close to the 1002
level. The weight of the above considerations favors the interpretation
that the targets were easier to detect for the diffuse light as compared
to the pattern mask conditions - given the qualifications as stated.

By whatever means one wishes to interpret the above data, the
results must be considered somewhat striking. For simply by removing
the contour information from the pattern mask stimulus, there occurred
what gay be a totally different phenomenon (although the negative inter-

action was not statistically significant). That is, the percentage decre-

 

ment produced by the flash for the 13:5 hemisphere condition was greater
than the right by 10.7%.

Most important for immediate purposes, however, are the answers to
the issues posed above. All predictions made for Experiment IV were con-
firmed. A "masking" effect was produced by the diffuse light stimulus;
the effect was (by the writer's reckoning) smaller than that produced by
the pattern; and the differential hemispheres X mask conditions interaction
effect predicted by the model for the pattern mask was not reproduced by
the diffuse light stimulus. (The interaction for Group V was significant
in the direction predicted at the 0.01 level, while that for Group VII
was not significant and showed a "trend" in the Opposite direction. Or
equivocally, the interactions were significantly different - p<:0.05).
Consequently, the hypothesis that the differential cross-field visual
masking phenomenon was produced by stray illumination is rejected. Simi-
larly, any "distraction of attention" type hypothesis is also rejected

since diffuse light flashing to either side of fixation must be considered

It .. I '-'i .‘.‘.}h‘Pm-uwwx-—“

 

79

equally distracting. The effect that 333 produced by the diffuse light
stimulus was likely due, at least partly, to stray illumination; but, since
a new interaction "trend" was observed, the possibility of involvement of
25525 cortical (hemispheric) mechanisms cannot be ruled out. Obviously,
statistical confirmation of this new "trend" must be established before
it is taken as a serious issue.

The obtained hemispheres X light flash interaction "trend" (if
confirmed) is indeed most difficult to account for in terms of a model like
the one presented here. To reiterate, the effect observed was that when

the stimulus to be verbally coded was flashed directly to the left hemis-

 

phere and the diffuse lighggto the righg, the masking effect was greater 1
than when the situation was reversed. A model which stresses hemispheric

differences in signal-to-noise ratios and relative stimulus arrival times

simply cannot explain that effect, at least not without the superimposition

of additional concepts.

At the moment, it seems that a model which £2213 account for the
above effect must include the following concept: the right hemisphere (not
necessarily the visual cortex) is somehow so specialized for the processing
of diffuse light that it produces a greater effect on pggh hemispheres, when
it is the recipient of the initial flash, than does the left hemisphere
when is is the initial recipient. In other words, on either the input or
the output side of the information processing equation, the right hemis-
phere may handle diffuse light in a different manner than does the left.

In the overall scheme of things, this is not so outlandish as the now accept-
ed notion that the left hemisphere "speaks" and the right does not. One
kind of evidence which supports the above possibility is the finding by

Jeeves and Dixon (1970) that the right hemisphere is at least a "more

80

efficient" processor of "unstructured" light stimuli than the left. They
measured motor (manual) reaction time to "unstructured" light flashed
alternately to the left and right hemispheres. After counterbalancing for
all possible combinations: nasal versus temporal hemiretinae, right versus
left hand, and left versus right hemispheres; they concluded that motor
reaction time was 5 to 8.5 ms faster to diffuse light flashed to the righg
as opposed to the left hemisphere. ?

Another, so far unmentioned, visual pathway is a possible mechanism
for interaction with diffuse light. This is the projection from the lateral F
geniculate through the superior colliculus and pulvinar structures to

"higher" visual centers in the temporal lobe (see Stone & Freeman, 1973).

 

This sub-cortical pathway is bilaterally represented and completely by-
passes the visual cortex. Although its real significance is so far unknown,
present evidence suggests that: it is phylogenetically an older pathway
(excluding the temporal lobe); it seems, in higher vertebrates, to be rela-
tively unrelated to form discrimination per se; it appears to have something
to do with the detection of moving objects; it may mediate the pupillary
response; and it responds easily to "light" versus "dark" as do some single
cells in the temporal lobe (Stone & Freeman, 1973).1 The fact that both
this pathway and the geniculo-striate path through the visual cortex appear
to terminate in the visual centers of the temporal lobe lends credence to

the possibility of a diffuse light masking effect in the temporal cortex.

 

lMy appreciation to Dr. J. I. Johnson, Michigan State University,
for the use of this reference chapter from a book still in press.

Chapter 6: GENERAL SUMMARY AND CONCLUSION

 

The purpose of this study was to develOp and test a new visual
masking paradigm which, if successful, could be employed as a procedure
for increasing observed laterality differences in visual perception. The
technique devised was to tachistosc0pica11y flash target information to
the "preferred" hemisphere while flashing a visual noise or masking pattern
to the hemisphere apposite. When the target information transfers from
the "preferred" to the apposite hemisphere via the corpus callosum, it was
predicted that processing of that information would be partially disrupted
in the "masked" hemisphere. In order to demonstrate that a gain in lateral-
ity difference was actually obtained, it was necessary to show that dis-
ruption of processing was greater in the "masked" hemisphere than it was in
the "preferred" hemisphere - which would also be influenced by mask informa-
tion crossing the callosum. In other words, when compared to no-mask con-
ditions, a difference in "percentage decrement" produced by the mask should
obtain between the "preferred" and "masked" hemispheres. A theoretical
model was developed which predicted that such a difference would be observed
on the basis of a higher "signal-to-noise" ratio that would obtain in the
"preferred" as compared to the "masked" hemisphere.

The specific goals of the study were to determine whether or not
such a phenomenon would occur and, if so, to study various stimulus para-
meters which would or would not influence the effect. The general findings
are summarized below.

81

82

l. Employing single alphabetic letters and the left hemisphere as
main processor, the predicted differential masking phenomenon was statistic-
ally confirmed.

2. Difference in energy levels between target and mask stimuli did
not substantially influence the masking effect.

3. Altering the stimulus onset asynchrony between target and mask
and therefore their arrival—arrival time relationships in the respective
hemispheres had a marked influence on the magnitude of the effect.

4. Size of the visual noise pattern was a crucial determinant of
the effect since masking nearly disappeared when the noise pattern was
reduced to l/lOth of its original area.

5. A difference in effect resulting from projection of the "small"
noise patterns to corresponding versus non-corresponding retinal areas
relative to the target could not be demonstrated.

6. Diffuse light, equal in brightness to that of the pattern mask,
did not produce the phenomenon predicted by the model and resulted in a
non-significant differential masking trend in a direction opposite to that
obtained for the pattern.

Results 1, 2, 3, 4, and 6 above supported in various ways the
hypothesis that the cross-field masking phenomenon was due to the inter-
action of central (i.e. cortical) rather than peripheral mechanisms.
Initial observation of the phenomenon itself (1 above) is argument for
the rejection of peripheral mechanisms since peripheral models cannot
account for the obtained laterality differences in masking effect. The
findings (2 and 3 above) that differing mask - target energy levels gig
pg; influence the phenomenon, whereas variation of the SOA parameter did,
agree with Turvey's (1973) conclusions regarding the parameters which

g2 and do not influence noise pattern masking by central mechanisms (via

83
the dichOptic procedure). Item 4 above indirectly supports the possibility
that "complex" cells in visual area 18 may have played a role in the mask-
ing effect because the mask size which proved an effective stimulus (com-
pared to the size which did not) corresponds to the receptive field dimen-
sions of complex cells in higher primates. Finally, the finding that
diffuse light did not produce the masking effect obtained for the pattern
permits rejection of the hypothesis that stray illumination within the eye
(the peripheral mechanism most likely to provide an interaction effect)
accounted for the phenomenon. In addition, it increases the tenability
of the notion that the effect obtained for the pattern mask involved an
interaction via the contour detectors in the visual cortex.

The combination of stimulus parameters which produced the largest
differential masking effect (Group V, Experiment II) was one which attempt-
ed to combine the hypothetical effects of differing signal-to-noise condi-
tions with a particular set of arrival-arrival time relationships between
target and mask in the respective hemispheres. Specifically, stimulus

onset asynchrony was adjusted so that simultaneipy of mask and target

 

arrival would occur in the low signal-to-noise hemisphere (the "masked"
hemisphere) and target would precede arrival of the mask in the high
signal-to-noise hemisphere (the "preferred" hemisphere). Actually, since
only verbal stimulus material was used, the left hemisphere was alternately
made to play the role of both the "masked" and the "preferred" hemispheres

by reversing target and mask stimulus fields. When the left hemisphere

was defined as "preferred" (target to left h. - mask to right h.), percent-
age decrement produced by the mask was only 8.51. But when the left hemis-
phere acted as the "masked" hemisphere (target to right h. - mask to left h.),
percentage decrement was 27.1% - a difference of 18.61. This effect was

significant at the 0.01 level (n=4).

84

In terms of the original goal to increase laterality differences
in perception of tachistosCOpically presented visual stimuli, the gain
produced by cross-field visual masking is somewhat small. For the best
set of stimulus conditions (just described) the EEELQ of the performance
of one hemisphere to that of the other increases by 251 under masking.
For example, if the ratio in, say, "recognition accuracy" for one hemis-
phere to the other is 2 to 1, then under cross-field masking it should
become 2.5 to 1. Similarly, a ratio of 4 to l improves to 5 to 1 under
masking - and so on. Obviously, if the ratio without maskppg is already
large (say, 3 to l or more), then employment of this paradigm seems
hardly worth the effort. On the other hand, for borderline instances,
i.e. ratios close to 1 to 1, employment Of this paradigm could make a
difference between detection and non-detection Of hemispheric differences
(assuming "true" differences do exist). As stated in the general intro-
duction to this paper, most left-right differences found for visually
presented stimuli 2;: close to borderline. Kimura (1973) summarizes in
ratio form some of the findings for visual stimuli. A left to right
hemisphere superiority is observed for letters (1.23 : l) and words
(1.47 : l); and a right to left superiority for two-dimensional point
location (1.18 : 1), dot and form enumeration (1.20 : 1), matching of
slanted lines (1.05 : 1), and stereoscOpic depth perception (1.28 : 1).
The average of these ratios of 1.24 to 1 should become 1.55 to 1 if cross-
field visual masking is used.

Whether the above gain produced by this masking paradigm is large
enough to aid the unilateral (left-right) study Of more complex informa-
tion processing models is a question that remains Open for the present.

There is, however, reason for Optimism. Chances are that the magnitude

85
of the above gain is a conservative estimate of what could be obtained for
more complex stimuli or tasks. The use of single alphabetic letters in
this experiment required a large number of observations (400 per subject)
to obtain reliability of measurement. (Even at that, many subjects did
not demonstrate a left hemisphere superiority at all under the no-mask con-
ditions.) This requirement enhanced the probability of extended practice
effects, and it has been shown (Schiller, 1965) that central masking with
a pattern stimulus considerably decreases with practice. Consequently, the
use of stimulus material requiring a small number of trials would seem to
favor observation of a larger overall masking effect, and possibly a larger
differential effect as well.

While it has been demonstrated that the cross-field masking tech-
nique is of value for increasing observed laterality differences in per-
ception, the model, or the set of interrelated explanatory constructs em-
ployed in the conceptual develOpment of the technique, has itself been sup-
ported by the results Of this study. The string of neurological assumptions,
though lengthy, has been shown to have the utility of generating behavioral
predictions in advance of the empirical observations - an essential criterion
for theory validation. Consequently, the potential of this technique as a
method for studying central neurophysiological processes in man should not
be overlooked.

It is well understood that direct study of the characteristics of
single neurons in the human cortex can be conducted only under the most
extreme of circumstances and even then involves difficult questions of
ethics. While the microelectrode studies on animals provide crucial leads,
less direct, behavioral, procedures are required for human subjects. The

"cross-field" as well as the dichoptic masking paradigms offer such possi-

bilities. By finding the size of a pattern which produces the optimum

86
cross-field masking effect (as suggested in Chapter 4), it is possible to
ipfgg the receptive field dimensions of single cells in areas 18 and 19
of the human visual cortex. Similarly, by systematically varying the
physical features of the masking pattern to produce an optimum interfer-
ence effect it is possible to infer the stimulus parameters to which the
cells are specialized. This same procedure might even be extended to the
study of "engram" coding by neurons in the visual area of the temporal
lobe.

Finally, brief mention will now be made of the need for some
further studies that bear on the theoretical issues of the masking tech-
nique itself. As mentioned above, this procedure should be tried with
more complex stimuli, and 4-1etter common nouns presented vertically would
be a good starting point. Secondly, if the proposed mechanisms are Operat~
ing, it ought to be possible to show that the effect works in the Opposite
direction. That is, stimuli should be chosen for which the right hemis-
phere has been shown to be the preferred processor. A differential mask-
ing effect should then be Obtained which is opposite in direction to the
one observed for this study. Thirdly, "target distance from fixation"
should be an interesting variable to investigate. The hypothetical mecha-
nisms proposed to be Operative for this paradigm are the ones which sup-
posedly relate to continuity of vision between the two half-fields and
binocular stereopsis along the vertical meridian. If that is the case,
then stimuli which project more than 3° lateral to the fixation point
should avoid this system entirely and would probably involve different com-
missural pathways. Finally, it should be shown that an effect is obtain-
able by totally eliminating any possibility for the occurrence Of peri-
pheral interactions. This would be made possible by combining the

dichoptic with the cross-field visual masking paradigms. By employment

87
of a 5-channel tachistoscOpe (or a 3-channel scope with polarizers) it is
possible to control both field of presentation and the eye to which it is
presented. For example, left-field presentations could be projected to
the temporal hemiretina of the right eye (and therefore to the right hemis-
phere) and right-field presentations to the temporal hemiretina of the lgfg

eye (left hemisphere); in other words, projection to geparate hemispheres

 

via corresponding retinae of the separate eyes. Obviously, in this instance,

 

only cortical mechanisms could interact. The effects of eye dominance
could be counterbalanced by making an equal number of cross-field presenta-
tions to the nasal hemiretinae of the two eyes which then project to the

contralateral hemispheres.

BIBLIOGRAPHY

BIBLIOGRAPHY

Bartley, S. H., and Fry, G. A. An indirect method for measuring stray
light within the human eye. J.0.S.A., 1934, 24, 342-347.

Bartley, S. H. The comparative distribution of light in the stimulus
and on the retina. J. Comp. Psychol., 1935, 12, 149-154.

 

Barton, M. J., Goodglass, H., and Shai, A. Differential recognition of
tachistoscOpically presented English and Hebrew words in right
and left visual fields. Percept. Mot. Skills, 1965, 21, 431-437.

 

Battersby, W. S., and Wagman, I. H. Neural limitations of visual excit-
ability. IV. Spatial determinants of retrochiasmal interaction.
American Journal of Physiology, 1962, 203, 359-365.

 

Berlucchi, G., Gazzaniga, M. 8., and Rizzolatti, G. Microelectrode
analysis of transfer of visual information by the corpus callosum.
Arch. ital. Biol., 1967, 105, 583-596.

 

Berlucchi, G., and Rizzolatti, G. Binocularly driven neurons in visual
cortex of split-chiasm cats. Science, 1968, 159, 308-310.

BishOp, P. 0., and Henry, G. H. Spatial Vision. Ann. Rev. Psychol.,
1971, 22, 119-160.

 

Bishop, P. 0., Henry, G. H., and Smith, C. J. Binocular interaction
fields of single units in the cat striate cortex. J. Physiol.,
1971, 216, 39-68.

 

Blakemore, C. Binocular depth perception and the Optic chiasm. Vis. Res.,
1970, 19, 43-47.

Boynton, R. M. Some temporal factors in vision. In W. A. Rosenblith (Ed.),
Sensory Communication. New York: Wiley, 1961.

BoyntOn, R. M., Enoch, J. M., and Bush, W. R. Physical measures of stray
light in excised eyes. J.0.S.A., 1954, 44, 879-886.

Bremer, F. NeurOphysiological correlates of mental unity. In J. C. Eccles
(Ed.), The brain and the unity of conscious experience. Cambridge:
University Press, 1965, p. 286.

 

van Buren, J. M. The retinal ganglion cell lsygr. Springfield, Ill.:
Charles C. Thomas, 1963.

 

Cragg, B. G., and Ainsworth, A. The tOpography of the afferent projections
in the circumstriate visual cortex Of the monkey studied by the
Nauta method. Via. Res., 1969, 2, 733-747.

88

89

Crovitz, H., and Zener, C. A group test for assessing hand and eye
dominance. Am. J. Psychol., 1962, 18 (2), 271-276.

 

Davis, H. PsychOphysiology of hearing and deafness. In S. S. Stevens
(Ed.), Handbook of experimental psycholggy. New York: Wiley,
1951.

Ditchburn, R. W., and Ginsborg, B. L. Involuntary eye movements during
fixation. J. Physiol., 1953, 119, 1-17.

 

Efron, R. The effect of handedness on the perception of simultaneity
and temporal order. Brain, 1963, 88, 261-284. (a)

Efron, R. The effect of stimulus intensity on the perception Of simul-
taneity in right- and left-handed subjects. Brain, 1963, 88,
285-294. (b)

Ettlinger, E. G. (Ed.) CIBA Foundation Study Group No. 20, Functions
of the corpus callosum. Boston: Little, Brown and Co., 1965.

 

 

Fox, C. A., Fischer, R. R., and Desalva, S. J. The distribution of the
anterior commissure in the monkey (Macaca Mulatta). J. Comp.
Neurol., 1948, 88, 245-278.

pus-..-:

Gazzaniga, M. S., Bogen, J. E., and Sperry, R. W. Observations on visual
perception after disconnexion of the cerebral hemispheres in man.
Brain, 1965, 88, 221-236.

Geschwind, N. Disconnexion syndromes in animals and man. Parts I and 11.
Brain, 1965, 88, 237-294 and 585-644.

Gordon, H. W., Bogen, J. E., and Sperry, R. W. Absence of the deconnexion
syndrome in two patients with partial section of the neocommis-
sures. Brain, 1971, 18, 327-336.

Grill, D. P. Variables influencing the mode of processing of complex
stimuli. Percept. & Psychophysics, 1971, 18 (1), 51-57.

Gross, C. G., Rocha-Miranda, C. E., and Bender, D. B. Visual prOperties
of neurons in inferotemporal cortex of the Macague. J. Neuro-
physiol., January, 1972, 88, NO. 1, 96-111.

Hays, W. L. Statistics for psychologists. New York: Holt, Rinehart, and
Winston, 1963.

 

Hecaen, H. Clinical symptomatology in right and left hemispheric lesions.
In V. B. Mountcastle (Ed.), Interhemispheric relations and cerebral
dominance. Baltimore: Johns HOpkins Press, 1962.

Heron, W. Perception as a function of retinal locus and attention. Am. J.
Psychol., 1957, 18, 38-48.

Hubel, D. H., and Wiesel, T. N. Receptive fields and functional archi-
tecture in two nonstriate visual areas (18 and 19) of the cat.
J. NeurOphysiOl., 1965, 88, 229-289.

 

 

9O

Hubel, D. H., and Wiesel, T. N. Cortical and callosal connections con-
cerned with the vertical meridian of visual fields in the cat.
J. NeurOphysiol., 1967, 88, 1561-1573.

Hubel, D. H., and Wiesel, T. N. Receptive fields and functional archi-
tecture of monkey striate cortex. J. Physiol., 1968, 195,
215-243.

Jeeves, M. A., and Dickson, N. F. Hemisphere differences in response
rates to visual stimuli. Psychonom. Sci., 1970, 88, 249-251.

Kahneman, D. Method, findings, and theory in studies of visual masking.
Psych. Bulletin, 1968, 18, 404-425.

 

Kimura, D. Cerebral dominance and the perception of verbal stimuli.
Can. J. Psychol., 1961, 18, 166-171.

Kimura, D. Left-right differences in the perception of melodies.
Q. J. Exp. Psychol., 1964, 18, 355-358.

Kimura, D. Dual functional asymmetry of the brain in visual perception.
NeurOpsychologia, 1966, 8, 275-285.

 

Kimura, D. The asymmetry of the human brain. Scientific American,
March, 1973, 70-78.

Levy, J. Possible basis for the evolution of lateral specialization of
the human brain. Nature, 1969, 224, 614-615.

Levy-Agresti, J., and Sperry, R. W. Differential perceptual capacities
in major and minor hemispheres. Proceeding: of the National
Academy of Sciences, 1968, 81, 1151.

Marg, E., Adams, J. E., and Rutkin, B. Receptive fields of cells in the
human visual cortex. Experientia, 1968, 88, 348-350.

McKeever, W. F., and Huling, M. D. Bilateral tachistosCOpic word recogni-
tion as a function of hemisphere stimulated and interhemispheric
transfer time. NeurOppychologia, 1971, 8, 281-288.

 

Milner, B., Branch, 0., and Rasmussen, T. Observations on cerebral domi-
nance. In A. V. S. de Rueck and M. O'Conner (Eds.), CIBA Founda-

tion's symposium on disorders of language. London: Churchill, 1964.

Mishkin, M. Visual mechanisms beyond the striate cortex. In R. Russell
(Ed.), Frontiers of Physiological Psycholggy. New York: Academic,
1966.

Mishkin, M., and Forgays, D. G. Word recognition as a function of retinal
locus. J. Exp. Psychol., 1952, 88, 43-48.

Mitchell, D. E., and Blakemore, C. Binocular depth perception and the
corpus callosum. Vis. Res., 1970, 18, 49-54.

~.. FLA .'-Im1
, .°
r”

,LEiJ' ~2- MJ ‘1 a 1;-

iii- 5.1.1.3.

91

Myers, J. L. Fundamentals of Experimental Desigp. Boston: Allyn and
Bacon, 1967.

Myers, R. E. Transmission of visual information within and between the
hemispheres: a behavioral study. In V. B. Mountcastle (Ed.),
Interhemispheric relations & cerebral dominance. Baltimore: Johns
HOpkins Press, 1962.

Orbach, J. Differential recognition of Hebrew and English words in right
and left visual fields as a function Of cerebral dominance and
reading habits. NeurOpsychologia, 1967, 8, 127-134.

 

Rizzolatti, G., Umilta, C., and Berlucchi, G. Opposite superiorities of
the right and left cerebral hemispheres in discriminative reaction
time to physiognomical and alphabetical material. Brain, 1971,
83, 431-442.

Robinson, D. A. The oculomotor control system: A review. Proc. IEEE,
1968, 88, 1032-1049.

Sanderson, K. J., Dorian-Smith, J., and BishOp, P. O. Binocular corre-
sponding receptive fields of single units in the cat dorsal
lateral geniculate nucleus. Vis. Res., 1969, 8, 1297-1303.

Schiller, P. H. MonOptic and dichOptic vision masking by patterns and
flashes. J. Exp. Psychol., 1965, 159, No. 2, 193-199.

Schiller, P. H. Behavioral and electrOphysiological studies of visual
masking. In K. N. Leibovic (Ed.), Information processing in the
nervous system. New York: Springer-Verlag, 1969.

 

Singer, W. Inhibitory binocular interaction in the lateral geniculate
body Of the cat. Brain Res., 1970, 18, 165-170.

Spehlmann, R. The averaged electrical responses to diffuse and to
patterned light in the human. EEG Clinical NeurOphysiology,
1965, 18, 560-569.

Sperry, R. W. Cerebral organization and behavior. Science, 1961, 188,
1749-1757.

Sperry, R. W. The great cerebral commissure. Scientific American, Janu-
ary, 1964, 42-52.

Sperry, R. W., and Levy, J. Mental capacities of the disconnected minor
hemisphere following commissurotomy. Paper presented at the
symposium on asymmetrical function of the human brain, annual con-
vention of the APA, Miami, Florida, September, 1970.

St. Cyr, G. J., and Fender, D. H. The interplay of drifts and flicks in
binocular fixation. Vis. Res., 1969, 8, 245-265.

Stone, J. The naso-temporal division of the cat's retina. J. Comp.
Neurol., 1966, 126, 585-599.

92

Stone, J., and Freeman Jr., R. B. NeurOphysiological mechanisms in the
visual discrimination of form. In Handbook of Sensory Physiology,
Vol. VII, Part 3. Berlin: Springer-Verlag, in press.

Thompson, R. F. Foundations of Physiological Psychology. New York:
Harper & Row, 1967.

Turvey, M. T. On peripheral and central processes in vision; inferences
from an information-processing analysis Of masking with patterned
stimuli. Psych. Review, 1973, 88, No. 1, 1-52.

Wilson, M. E., and Cragg, B. G. Projections from lateral geniculate
nucleus in the cat and monkey. J. Anat., 1967, 101, 677-692.

White, M. J. Laterality differences in perception: A review. Psych.
Bulletin, 1969, 18, No. 6, 387-405.

Zeki, S. M. Interhemispheric connections of prestriate cortex in monkey.
Brain Res., 1970, 18, 63-75.

 

Zeki, S. M. Representation of central visual fields in prestriate cortex
of monkey. Brain Res., 1969, 18, 271-291.

 

APPENDICES

93

APPENDIX A

Brief Review of Controversial Issues Concerning Left Versus Right Field
Differences in Tachistoscopically Presented Visual Stimuli

Beginning with Mishkin and Forgays'(l952) and Heron's (1957)
lateral presentation experiments, a large number of such studies have been
conducted (see review by White, 1969), mostly with verbal stimuli. The
general findings are that for unilateral tachistoscOpic presentation of
verbal stimuli, a right field (left hemisphere) superiority is demonstrated.
However, with bilateral presentations, a left field superiority, under some
stimulus conditions, seems to predominate. This has given rise to the
"reading scan hypothesis" which provides a parsimonious explanation for
both observations. Briefly, it states that for unilateral presentation the
subject tends to scan (from the fixation point) the neural trace from a
left to right direction according to his reading habits, and for bilateral
presentation or presentation of material across the vertical meridian, the
subject starts "reading" from the left-most point. McKeever and Huling
(1971) who find a right field superiority under bilateral conditions, have
countered the reading scan hypothesis by arguing that some of the early
studies did not provide adequate control of fixation and by citing recent
experiments by Barton et a1. (1965) and Orbach (1967) who found right
field superiority with native Israeli subjects (who read from right to
left) and tachistoscopic presentation of Hebrew words. Perhaps the fair-
est assessment is provided in White's (1969) conclusion where he categor-
izes into two sets the experimental conditions of the conflicting studies:
(a) those in which the exposure durations were well above threshold,

50 - 150 ms (which support the reading scan hypothesis) and (b) those in
which displays were presented at threshold durations (which support the
lateral specialization hypothesis). White states: "It is to be expected
that as stimulus information and exposure durations tend towards the mini-
mal, these factors (hemispheric, etc.) should dominate a "scanning" pro-
cess which itself demands maximal information and time to process it."
The present project was concerned only with the second (b above) type of
stimulus condition.

 

94

APPENDIX B

Errors (Guesses) Classified as "Stimulus Dependent" for Each Target Letter
Used in Experiment I (Error Analyses 1 and 2)

 

 

Error Analysis 1 Error Analysis 2
Actual Letters guessed Actual Letters guessed
target (errors)i target (errorg)

A - V, H A -

B - E, S, R, P B - S

C - G, O C - O

D - O D - 0

E - F, B, E - F

F - E, P, R F - E, P

G - C, 0 G - 0

H - M, W, N, A, R H - M, W

I - T, L, I - T, L

J - U, L, J - U

K - 11,): K - 8.x

L - J, I L - 1

M - N, W, H M - H

N - M, H, W N -

0 - D, C, G, U 0 - C, G, D

P - F, R, B P - R, F

R - K, P, B, H, F R - P, K

S - B, 2, E S - B

T - I T - I

U - V, J, W, O U - J

V - U, W, Y, X V -

W - H, M, N, U W - H

X - Y, V, X X -

Y - X, V y -

Z - S 2 -

Summary Table and Analysis of Variance for Error Analysis 1 Based on

95

APPENDIX C

Percentage of Stimulus-Dependent Errors as the Dependent Variable

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

average

diff. in mean T exp.

1 decr.

across H y(RH-LH)

time (ms) @
threshold

 

Table 16. Summary table for error analysis 1

T to T to

left H right H
Group I nO-mask 57.9 38.3
M=2.50 ft.l. mask 42.6 38.6
T=0.40 ft.1. Z decr. 26.4 -0.8
Group II no-mask 39.2 48.4
M=1.05 ft.1. mask 40.3 34.3
T=O.40 ft.l. Z decr. -2.8 29.1
Group III no-mask 56.9 44.3
M-2.50 ft.l. mask 40.8 30.2
T30.20 ft.l. Z decr. 28.3 31.8
Group IV nO-mask 46.1 39.2
M-2.50 ft.1. mask 34.0 30.4
T-0.10 ft.1. Z decr. 26.2 22.4
Note.

decrement produced by mask

Table 17.

Source

Total

Analysis of variance for error analysis 1

Between subjects (S)

Groupe (A)

 

S/A

Within subjects
Hemispheres (B)

AB

SB/A

Mask conditions (C)

89
SC/A
1_3_<_:_
5.3.9.
SBC/A

SS

7068—.3 14

1721.
448.
1272.
5346.
731.
483.
881.
1555.
181.
797.
8.
481.
226.

671
904
767
643
026
502
612
316
309
162
629
999
087

48.1
40.6
12.8

MS

149.
106.

731.
161.

73

635
064

026
167

.467
1555.
60.
66.
8.
160.
18.

316
436
430
629
666
841

-27.2

31.9

3.5

-3.8

Pa

1.411 (ns)

4.9

4.4

8.3

10.6

M=mask; T'tsrget; H=hemisphere; R-right; L=left; 1 decr.=percentage

9.950 (p < 0.005)

2.194 (ns)
23.413

0.910 (ns)
0.458
8.528

(ns)

(p< 0.001)

(p < 0.005)

96

APPENDIX D

Summary Table and Analysis of Variance for Error Analysis 2 Based on
Percentage of Stimulus-Dependent Errors as the Dependent Variable

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 18. Summary table for error analysis 2
diff. in mean T exp.
T to T to average 2 decr. time (ms) @
left H right H across H (RH-LH), threshold
Group I no-mask 31.9 28.0 30.0
M'2.50 ft.l. mask 25.2 21.7 23.5 1.5 4.9
T=0.40 ft.l. 1 decr. 21.0 22.5 21.8
Group II no-mask 23.5 30.2 26.9
M=l.05 ft.1. mask 26.9 17.6 22.3 56.2 4.4
T=0.40 ft.l. Z decr. -14.5 41.7 13.6
Group III nO-mask 30.6 25.4 28.0
M82.50 ft.1. mask 25.4 18.0 21.7 12.1 8.3
T=0.20 ft.l. Z decr. 17.0 29.1 23.1
Group IV no-mask 33.0 21.3 27.2
M=2.50 ft.l. mask 17.4 16.0 16.7 -22.4 10.6
T=0.10 ft.l. 1 decr. 47.3 24.9 36 1

 

 

 

 

Note. M=mask; T=target; H=hemisphere; R'right; L=left; Z decr.-percentage
decrement produced by mask

Table 19. Analysis of variance for error analysis 2
Source 88 88’ 88 F

Total 63 3529.04 —

Between subjects (S) 15 901.44

Groups (5), 3 171.21 57.07 0.93 (ns)

S/A 12 730.23 60.85

Within subjects 48 2627.60

Hemispheres (B) 1 331.70 331.70 26.56 (p<10.001)

88 3 78.40 26.13 2.09 (ns)

SB/A 12 305.92 12.49

Mask conditions (C) 1 754.19 754.19 22.88 (P< 0.001)

88 3 62.34 20.78 0.63 (ns)

SC/A 12 395.58 32.97

pg 1 16.50 16.50 0.57 (ns)

_A_B_Q 3 336.79 112.26 3.89 (p<0.05)

SBC/A 12 346.18 28.84

   

"'cll'il'fill‘l‘iflllfll'zliflljtlfijjflflflifljfllflfifififl