J

IWIH‘J‘I

w
\
m

WWW

—{(Q_\
CDCO-b

 

THESIS

I)ate

This is to certify that the

thesis entitled

DIRECTED RESPONSE LEARNING BY
OBSERVATION OF AN APPETITIVE
PAVLOVIAN CONTINGENCY

presented by

CARLA FLORENCE CLOS

has been accepted towards fulfillment
of the requirements for

M.A. degree in PsychoTogy

_, / 1/
. Major pro sor
/ I

12/9/82

 

0-7639

MS U is an Affirmative Action/Equal Opportunity Institution

ifiifié’ffi E: Y

a? w . ‘
. Mfimaagmm E mite

Um'msessfi try

 

aw'i’"

W l

DIRECTED RESPONSE LEARNING BY OBSERVATION OF
AN APPETITIVE PAVLOVIAN CONTINGENCY

By

Caria Fiorence C105

A THESIS

Submitted to
Michigan State University
in partia] fquiTTment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychoiogy

1983

/3’7—250l-

ABSTRACT

DIRECTED RESPONSE LEARNING BY OBSERVATION OF
AN APPETITIVE PAVLOVIAN CONTINGENCY

By

CarTa FTorence CTos

Three experiments expiored modeT-absent observationaT Tearning of
barpressing in rats. The observation of two mechanicaTTy activated bars
(5+ and 8-) was foTTowed by a transfer of training test. In Experiment
1 the contribution of pTace Tearning was assessed by removing position
cues. Mates tested before reversaT of observational Tearning performed
pooriy. MaTes reversed before tested pressed the S+ bar 83% of the
time. Habituation of expToratory responses was invoked to account for
this difference. FemaTes were run in Experiment 2. FemaTes performed
poorTy on test, but extended training improved their performance. The
poor performance of femaTes compared with maTes suggested that gonadaT
steroids pTay a role. In Experiment 3, amount of magazine training
was manipuTated. Conditioning to the 3+ was retarded with 100

magazine triaTs, as was directed responding to 5+ (context-biocking).

To my mother,
who inspired and rewarded inquisitiveness

11

ACKNOWLEDGMENTS

I wish to express my gratitude to Dr. M. Ray Denny who
contributed so much to aTT phases of this thesis. My association
with Dr. Denny has provided me with a rare opportunity to receive
persona], considerate, and schoiariy guidance.

In addition, I woqu Tike to extend my sincere appreciation to
the other members of my committee, Dr. M. RiTTing and Dr. L. Hyman.

And, finaTTy, a word of thanks to my husband, Tom, for his

unfaiiing morai support.

TABLE OF CONTENTS

Page

LIST OF TABLES ........................ v
LIST OF FIGURES ........................ vi
INTRODUCTION . . ....................... I
EXPERIMENT I ......................... 17
Method ......................... 18
Subjects ....................... 18
Apparatus ....................... T8
Procedure ....................... T9

ResuTts and Discussion .................. 21
EXPERIMENT 2 ......................... 28
Method .......................... 29
Subjects ....................... 29
Apparatus ....................... 29
Procedure ....................... 3O

ResuTts and Discussion .................. 3O
EXPERIMENT 3 ......................... 38
Method .......................... 40
Subjects ....................... 40
Apparatus ....................... 41
Procedure ....................... 4T

ResuTts and Discussion .................. 42

LIST OF REFERENCES ...................... 48

iv

Tab]e

LIST OF TABLES

OutTine of the Experimenta] Phases for the Two Groups
in Denny, Be]], and C105 Observationa] Learning

Procedure .........................

Data from the five phases of the observationa] ]earning
study in Denny, Be]], and C105 (Note 1), number of
tria]s to observationa] ]earning criterion (inc]uding
criteria] day) and percentage barpress responses to

stimu]i over 10 test tria]s are ]isted ..........

Data from the five phases of Experiment ]. Number of
tria]s to observationa] ]earning criterion (inc]uding
criteria] day) and percentage barpress response to
stimu]i over ]0 test tria]s are ]isted .......

Data from five phases of Experiment 2. Number of tria]s
to discrimination criterion (inc]uding criteria] day)
and percentage barpress response to stimu]i over 10

test tria]s are ]isted ...............

Performance of Subjects Grouped by Number of
Pretraining Tria]s .................

Page

]2

T3

23

3]

32

LIST OF FIGURES

Figure Page

]. The two-choice observation apparatus (A) The Tight
above right bar. (B) Right bar. (C) Wooden b]ock
be]ow right bar (removed for the present series of
experiments). (D) Left bar. (E) Hardward c]oth tunne]-
door in up position, when rat can ]eave or enter tunne].
Door is down when rat is in tunne] watching bars being
activated. (F) Hardward c]oth tunne] in which rat ob-
serves bars and turns around to obtain food reward.
(G) Food tray in which pe]]et is dropped through tube. . . 9

2. Mean percent responses on CS+ and CS- tria]s for each
of the four groups in Experiment 3. The n_represents
number of subjects used in ca]cu]ating the mean
percent correct for that b]ock. The arrows indicate
when subjects reached overaT] criterion. Subjects
reaching criterion within a 40 tria] b]ock were
inc]uded in ca]cu]ating the mean for that b]ock ...... 44

3. Mean percent correct response on test as a function

of the number of pretraining food-onTy tria]s. Note
abscissa is p]otted on a ]ogarithmic sca]e ........ 45

vi

INTRODUCTION

Over the years it has become increasing]y c]ear that the response-
reinforcer contingency is not necessary for ]earning a response (Lajoie
& Bindra, ]976). As these investigators point out, "phenomena of response
substitution (e.g., Lash]ey & McCarthy, ]926), observationa] ]earning
(e.g., Kohn & Dennis, ]972), and ]earning without responding (e.g.,
Dodwe]] & Bessant, I960; So]omon & Turner, 1962) demonstrate that a new
response may be acquired without its occurrence during the training
tria]s.”

As Browne (]976) has pointed out,

The question of whether subjects can ]earn mere]y by
watching, without acting,is reTated to a persistent
confTict in theories of anima] ]earning, the question

of what is ]earned. According to cognitive theories,
]earning is mainiy a matter of gaining know]edge about
reTationships between events. On the other hand, accord-
ing to stimu]us-response theories, ]earning is primari]y
a matter of forming associations between stimu]i and
responses. During the 1930's and 1940's, the va]idity of
the two views was debated furious]y, but inconc]usive]y.
Partia]]y as a reaction against the inconc]usive

nature of the debates and partia]]y due to the strong
anti-theoretica] inf]uence of Skinner, issues surround-
ing the question of what is ]earned were avoided during
the 1960's. (p. 287)

The arguement continues today (Davey, Oak]ey & C]e]and, 1981; Locurto,
Duncan, Terrace & Gibbon, 1980; Me]]gren & O]son, 1980).

2

Various theoretica] schemes have been proposed to accommodate the
facts and investigate the issue of what is ]earned in a given situation.
Estes (]969, ]972) has suggested, essentia]]y, that three types of
associative ]inkages (contingencies) are the basis of ]earning:
stimu]us—stimu]us (S - 8*), stimu]us-response (S - R), and response-
stimqus (R - S*) contingencies; with each type of contingency being
sufficient in itse]f to support certain behaviora] modifications. Going
beyond Estes, Na]ker (1969) accepted the possibi]ity of response-response
(R - R) ]inkages. Bo]]es (]972) takes the position that behavior modi-
fication may arise from either response-stimu]us (R - 5*), or stimu]us-
stimu]us (S - S*) expectancies. Either type of contingency is suffi-
cient, but neither is necessary. Denny (]967) wou]d argue that the
consistent pairing of the stimu]us and the response is both necessary
and sufficient for ]earning. In contrast, Bindra (]978) c]aims that
a]] ]earned behaviors are eprained by S - S* associations, denying any
expTanatory importance to a response-reinforcer association. Within
Bindra's framework, the S - S* contingencies_are both necessary and
for a]] instances of instrumenta] ]earning.

The observationa] ]earning procedure is particu]ar]y we]] suited to
assessing the importance of the associations these theoretica] schemes
stress as it was designed to determine whether ]earning can occur
through mere exposure to a stimu]us sequence, that is, in the absence of
the performance of the to-be ]earned response and response contingent
reinforcement. In the typica] appetitive observationa] ]earning situ-

ation, i

an experimenta] subject is a]]owed to observe the perfor-
mance of a task by a demonstrator or mode] of the same
species. Nhi]e the subject is observing, he has no chance
to perform the observed behavior and receives no direct

3

reinforcement. After observation, the mode] is removed
from the experimenta] situation, and the observer

is tested on the observed task. If the observer ]earns
the task faster than contro] subjects that have not
observed a mode], it is assumed that the observer
]earned something about the task during observation that
faci]itated subsequent acquisition. ( De] Russo,1975,
p. 76

In the first part of the century, studies were conducted to deter-
mine if anima]s cou]d ]earn by observation (Berry, 1906, ]908; Haggerty,
]909; Thorndike, 19]]; Watson, ]908). The issue remained unreso]ved, as
both positive and negative resu]ts were reported. It has been suggested
that the discrepancy in resu]ts characterizing these ear]ier experiments
was due to the poor experimenta] methods and ]ack of proper contro] groups.
Warden and Jackson (]935) set up strict criteria for observationa]
]earning which e]iminated ear]ier confounds such as the possibi]ity for
tria] and error ]earning. Their experiment tested rhesus monkeys on ob-
servationa] ]earning of four tasks: pu]]ing a chain, manipUTating a knob,
operating a ]atch, and operating a doub]e ]atch. The experimenters found
that observers performed significant]y better than contro] subjects and
conc]uded that monkeys can ]earn by observation and, in fact, exhibit a
strong tendency to imitate. Observationa] ]earning has been demonstrated
in a number of species and in a variety of ]earning situations since
Warden and Jackson's experiment (Corson, ]967; De] Russo, 1971; Dubanoski
& Parton, 1971; Hayes & Hayes, ]952; Herbert & Harsh, ]944; Jacoby &
Dawson, 1969; Myers, ]970).

Whi]e evidence indicates that passive watching faci]itates subsequent
performance of a task, ]itt]e is known about the properties of the
faci]itating effects of the appetitive observationa] experience. More

recent]y, however, experimenters have turned their attention from mere

4
demonstration to a c]oser scrutiny of the observationa] ]earning processes
itse]f (DeT Russo, 1975; Groesbeck & Duerfe]dt, ]97]; Kohn & Dennis, 1972).
Three stimu]us cues can be identified in the standard appetitive obser-
vationa] ]earning procedure: the manipuIandum, some movement or change
in the manipuTandum, and the mode]. The effect of these variabTes might
be better specified if each stimu]us were presented in a manner a]]owing
the separate effects of each cue to be exp]ored. It may be, as we sha]]
argue, that the mode] is re]ative]y unimportant for faci]itating ob-
server performance of the task aside from its ro]e in activating the
manipuTandum or in maintaining the observer anima]'s attention on each
observationa] tria]. In such a capacity, the mode] wou]d serve simpTy
as a means to an end in the subject's observation of the consistent
pairing of stimu]us and reinforcer. In other words, the observer
anima] gains information about the environmenta] contingencies (]earns
S - S re]ationships) apart from any mode]ing (imitating) component in
the observationa] situation.

The re]ative contributions of the informationa] and the mode]ing
components present in the standard appetitive observationa] ]earning
situation were examined in a study conducted by Groesbeck and Duerfe]dt
(]97]). Each of six groups of ]0 rats received observationa] ]earning
tria]s of a visua] discriminating task in an e]evated Y maze. Subjects
were then tested on the task for the number of tria]s and errors to
reach criterion (]0 consecutive correct tria]s). Various combinations
of observationa] ]earning components (mode]ing/imitating, informationa]
vicarious reinforcement, fo]]owing) were avai]ab]e to the six groups.
For the information onTy (mode]-absent) group, observationa] ]earning

tria]s invo]ved the experimenters knocking over the S+ pane] to revea]

5
the water bottTe at the end of the correct arm of the maze. Mode]-
present groups observed demonstrator rats knock the S+ pane] over and
proceed down the correct arm of the maze to the water bott]e. No
differences were found in terms of the number of errors prior to reaching
criterion between the group having information on]y and the group having
both mode]ing and information components. However, the resu]ts did
indicate that, in terms of the number of tria]s to reach criterion,
the groups with the mode] present ]earned significant]y faster than did
groups without the mode]. The researchers conc]uded that mode]ing of
the demonstrator's performance (copying the mode]'s behavior) was the
most important variab]e in the observationa] ]earning situation. A
c]oser ]ook at the data, however, indicates this conc]usion may have
been premature. No differences were found between the mode]-present
and the modeT-absent groups in the mean number of tria]s unti] the
first correct discrimination choice was made. Differences between
groups on]y appeared with extended training to reach the criterion of
]0 consecutive correct tria]s. The difference between groups that
showed up with extended training cou]d have ref]ected the observers'
paying better attention in the standard condition rather than actuaTTy
copying what the mode] did. Attention in ear]y tria]s, presumab]y,
was mediated by orienting responses to the nove] situation and therefore
approximate]y equa] across groups. As training proceeded, however,
the mode] cou]d have served to mediate consistent attention to (and thus
observation of) the S - S contingencies --the pane]s paired with water.
In contrast, the mode]-absent groups may have ceased to observe further
S - S pairings after the nove]ty of the situation ceased to sustain

attention. Since no mention was made of any method of verifying that

6

subjects in any of the groups actua]]y observed the pane]-water pairings,

 

differences in attention remain as a possibTe confound. So, as yet,
the issue of the re]ative importance of the various components in
observationa] ]earning is far from conc]usive. Indeed, evidence from
research with humans seems to support the assertion that the mode] may
be re]ative]y unimportant aside from insuring that the subject observe
the consistent pairing of the stimu]us and reinforcer. Jacobson and
Sisemore (]976) reported that the observation of contingencies
(mechanica] bar depression paired with reinforcement) was sufficient
to e]icit a fu]] b]own response in subsequent test tria]s.

If the mode] is re]ative]y unimportant, then appetitive obser-
vationa] ]earning cou]d be reana]yzed as an instance of a c]assicaTTy
conditioned response system. The faci]itation of performance of a task
from the observationa] ]earning procedure may be derived from the
deve]opment of behavior directed toward a stimu]us that, by virtue of
its contingent re]ation to the reinforcer, serves as a signa] of
the forthcoming reinforcer. That behavior can be steered toward a
stimu]us that predicts the imminent arriva] of a reinforcer is we]]
documented in the ]iterature as instances of autoshaping or signtracking
(see Hearst & Jenkins, ]974, for a review).

Observationa] ]earning, from this perspective (mode]-absent),
resemb]es autoshaping studies where the subject is b]ocked or restrained
from approaching the CS and/or the US during a]] pairings of the two
stimu]i (Browne, ]976). Prevention of such directed actions may be
accomp]ished, for examp]e, by inserting a barrier between the subject and
the CS or US ]ocation, or by harnessing the subject in a p]ace from

which CS or US presentations are visibTe but inaccessib]e. If actua]

 

7
occurrences of approach and contact are not crucia] for thedeve]opment
of tendencies to approach and contact the CS, then prior "observation"
of a positive contingency between CS and US shoqu produce immediate or
rapid appearance of approach and contact responses when the CS is fina]]y
made accessibTe to the subject (Hearst, ]978).

Kirby, Muir, and Moore (cited in Moore, ]973) exposed pigeons to
standard autoshaping procedures, except that a hardware c]oth screen
prevented the birds from access to the key. The birds were a]]owed
access to grain when the hopper was raised. The birds observed 60
CS+ tria]s (white key]ight fo]]owed by grain) and 60 CS- tria]s (vertica]
stripes on a key fo]]owed by no grain) for 20 dai]y sessions. The
pigeons pecked the CS+ the very first time it appeared during a test
with the key avai]ab]e. AT] 2]4 pecks that occurred during the test
phase went unreinforced. Thus it wou]d appear that ]earning of the
stimu]us-reinforcer re]ation had taken p]ace in the absence of pro-
gressiveTy c]oser approach movements and that autoshaping and re]ated
phenomena invo]ving observation of contingencies cou]d not be expTained
in terms of adventitious reinforcement of keypecking. However, Wesse]]s
(1974) pointed out that specific directed movements were c]earTy
conditioned in the observation phase of the experiment. The pigeons
repeated]y directed their beaks into the screen and made biting move—
ments. It is important that the directed movements emp]oyed as the in-
dex of the stimu]us-stimu]us ]earning not be conditioned in the
observation phase of the experiment, or e]se there is no c]ear
separation of the associative ]inkages that ]ed to the behavior

modification. Despite such difficu]ties with these experiments, they

 

 

 

8
have been offered as evidenace of the importance of a stimu]us-stimu]us
contingency as a basis for certain behavior modifications.

Research performed by Denny, Be]], and C]os (Note ]) attempted to
address the above issues and contro] for some of the confounds that
]ed to the difficu]ty of making c]ear-cut determinations of the con-
tingencies operating in each phase of a ]earning situation invo]ving
observation of environmenta] events. First, measures were taken to
insure that anima]s were indeed observing the environmenta] events.
A]so, there were no overt directed movements during the observation
phase that corresponded to anything ]ike the behavior which served as
the measure of stimu]us-stimu]us ]earning. Furthermore, the study
addressed not on]y the issue of the initia] acquisition of a directed
contact response, but a]so the subsequent persistence and strength of
these responses despite changes in the reinforcement context in which
the responses occurred (the ]atter issue is often addressed in studies
of the omission effect). The Denny, Be]], and C]os study (Note ])
wi]] be discussed here in some detai] as the present series of
experiments extended their ana]ysis of the observationa] ]earning
procedure.

Their experiment cou]d be c]assified as either a b]ocked discrim-
inative autoshaping study or a modeT-absent observationa] ]earning study,
since both share essentiaTTy the same e]ements--an observation of con-
tingencies phase and a transfer of training testing phase. In the
observation phase, the subjects were required to observe the
activation of either the right or ]eft bar and its attendant stimu]i
(]ight and/or noise), fo]]owed by the deTivery of an accessibTe food

pe]]et in the food tray (see Figure 1). The food pe]]et was not

.onzp cmsogcp nwaaogv

mw prqu sows: cw xmgu woo; va .vgmzmg noow :wwpno op uczogm mcgzp ecu mgmn mw>gwmno um;
;o_;3 cw chczp spoFo mngvgm: Adv .vam>vpom mcwmn mgmn mcwzopwz Focczw cw mw pm; cog: cacv
mw Loco .Foccsp Lapse Lo m>wm_ can we; cos: .cOEpwmoa a: c? Loov-_w::=p sue—u mgmzuxmx Amv
.Lmn awe; “av .AmpcwEwLmaxm we meme ucwmwga mzp Low Uw>oEva Ema “saw; 30—wa xoo_n

:wnooz Auv .Lcn pcmwm nmv .Lmn “saw; m>onm pcm?F one A<v mspmemaaw cowum>gwmno mowoco-ozp one

Eomsm

._ wgzmwd

 

 

]0
de]ivered unti] the rat had jack-knifed around in the tunne] en route
to the food tray. This procedure insured that the moving bars and
attendant stimu]i maintained exc]usive stimu]us contro] of the go/no-
go behavior. Activation of the other bar (the CS-) was fo]]owed by a
period of nonreinforcement, and jack-knifing around in the tunne] had
no scheduied effects except possib]y to increase the intertria] inter-
va] (ITI) if the subject fai]ed to return to the tunne] door by the
end of the previous ITI. Tria]s were administered in b]ocks of 20 with
one or two b]ocks being presented dai]y unti] the discrimination
criterion was reached. The criterion was (a) inhibiting on 9 out of
]0 S— tria]s in a b]ock by continuing to face the test chamber for at
]east ]0 seconds after the 5- bar moved, and (b) ]eaving the tunne]
door for the food tray in ]ess than ]0 seconds on ]0 out of ]0 S+
tria]s in a b]ock. Accurate performance on this go/no-go discrimination
was thus used as an index of the ]earning of the S - S re]ationships. The
number of tria]s to criterion inc]uded the criteria] day.

Once the discrmination criterion was met,the tunne] door was
raised and the subject was a]]owed to enter the test chamber with the
tunne] door c]osed behind it. If the subject pressed a bar it was
removed from the chamber and returned to its home cage. Thus, with
respect to the food tray the subject was neither reinforced nor non-
reinforced for barpressing. If the rat did not press a bar within
five minutes it was removed from the test chamber and returned to its
home cage. On each of three succeeding days (test days 2, 3, 4) the
rat was placed in the tunne] and given one b]ock of 20 booster tria]s
fo]]owed immediater by a test as described. On test days 5, 6, and 7

the rat was tested both prior to and fo]]owing the b]ock of 20 tria]s.

]]
This gave a tota] of 10 test tria]s over a 7 day period. Fo]]owing the
7 test days these subjects received further observationa] training with
cues reversed. Upon reaching the same criterion as used for the
origina] discrimination the subjects were again given ]0 test tria]s
over 7 days.

A second group of subjects received identica] treatment as this
first group, Group TBR (Tested Before Reversed), unti] the origina]
discrimination criterion was met. Upon reaching criterion, however,
the second group was reversed and given more observationa] training
tria]s. When the second group, Group RBT (Reversed Before Testing),
reached reversa] discrimination criterion, it began the testing phase
as outTined above. Fo]]owing the 7 test days, these subjects received
further observationa] training with the cues reversed a second time.

A fina] testing phase ended the experiment for these subjects.
Whether the subjects were a]]owed access to the CS bar in the testing
phase prior to or fo]]owing reversa] of the S - S observationa]
]earning in the tunne] was the critica] difference between the two
groups (see Tab]e ]).

A]] of the subjects in Group TBR after a brief exp]oration period
pressed the CS+ bar on the initia] test tria], and continued pressing
the CS+ bar on 87% of the ]0 test tria]s. These resu]ts, presented
in Tab]e 2, are consistent with findings of Browne (]976) and other
researchers who conc]ude that the initia] acquisition of directed
movements depend main]y on stimu]us—reinforcer correTations. When the
subjects from this group were tested, Phase 5, after receiving S - S
reversa] training in the tunne], not one subject pressed the new S+

bar on the initia] test tria]. As a group, they se]ected the origina]

Tab]e ]

Out]ine of the Experimenta] Phases for the Two Groups in

Denny, Be]], and C]os Observationa] Learning Procedure

 

 

 

 

 

Group Phase
] 2 3 4 5
TBR Observation Test Observation Test
of cues of reversed
cues
RBT Observation Observation Test Observation Test
of cues of reversed of reversed
cues cues
TBR Tested Before Reversed

RBT

Reversed Before Tested

]3

Tab]e 2

Data from the five phases of the observationa] ]earning study in Denny, Be]], and
C]os (Note ]). number of tria]s to observationa] ]earning criterion (inc]uding
criteria] day) and percentage barpress responses to stimu]i over ]0 test tria]s

 

 

 

 

are ]isted
Group Subject Phase
1 2 3 4 5
TBR R 110 (3+) 246 - (3+) 70% (BL+) 220 (3-) 50%
(BL—) 30% (BL+) 50%
R 222 (3+) 100 - (3+) 90% (BL+) 200 (3-) 30%
(BL-) 10% (BL+) 20%
R 666 (3+) ]60 - (3+) 100% (BL+) 157 (3-) 100%
(BL-) 0% (BL+) 0%
R 010 (3L+) 308 - (3-) 0% (3+) 140 (3+) 20%
(BL+) 100% (BL—) 30%
R 111 (BL+) 190 - (3-) 30% (3+) 316 (3+) 10%
(BL+) 70% (BL-) 90%
R 333 (BL-) 130 - (3-) 10% (3+) 120 (3+) 10%
(BL+) 90% (BL-) 90%
Mean 197.3 - (+) 36.7% 192.2 (+) 13.3
(-) 13.3% (—) 81.7
3.0. 71.9 - 71.2
RBT R 003 (BL+) 130 (3+) 172 (3+) 90% (3L+) 237 (3-) 10 %
(BL-) 10% (BL+) 0%
R 001 (3L+) 100 (3+) 160 (3+) 90% (3L+) 140 (3-) 90%
(BL-) 10% (3L+) 10%
R 005 (3L+) 389 (3+) 310 (3+) 100% (BL+) 340 (3-) 100%
(BL-) 0% (31+) 0%
R 100 (3+) 307 (3L+) 233 (3-) 10% (3+) 30 (3+) 0%
(BL+) 90% (BL-) 100%
R 004 (3+) 230 (BL+) 213 (3-) 10% (3+) ]86 (3+) 10%
(BL+) 90% (BL-) 90%
R 002a (3+) 200 (3L+) 30 (3-) 0% - -
(BL+) 0%
Mean 241.2 228.6 (+) 92% 196.6 (+) 4% ,
(-) 3% (-) 96%
5.0 111.3 67.7 93.9

 

Note.

 

Parentheses indicate training and test stimu]i in each phase for that subject.

“TBR = Tested Before Reversed; RBT = Reversed Before Tested; 8 = Bar on]y stimu]us
(]eft stimu]us);BL = Bar and ]ight stimu]us (right stimu]us).

aSubject was a non]earner (never performed on test) and was dropped from experiment.
Data was not inc]uded in mean or 5.0. for group.

]4

S+ bar (now the S-bar) 0n 8].7% of the ]0 test tria]s. Group TBR did
not reverse on test. Group RBT showed reversa] ]earning on test (Phase
3) with 5 out of 6 rats in this group reversing and se]ecting the new
S+ bar on 92% of the ]0 test tria]s. With Fisher's Exact Test the
difference between the groups was high]y significant (p_< 0.00]).
However, when Group RBT was given S - S reversa] training in the tunne]
a second time, the subjects pressed the new S+ bar (in Phase 5) on on]y
4% 0f the ]0 test tria]s. Group RBT did not reverse a second time.
After both groups had been tested once, by Phase 5, the two groups
were showing simi]ar performance on test (i.e., not reversing).

According to e]icitation theory, (Denny, ]97]), reversa] ]earning
is accomp]ished by the extinction of the origina] response through the
]earning of a competing response in the same stimu]us situation. The
fai]ure to reverse after testing suggested that the origina] S - R
]earning (consistent pressing of a particu]ar bar during the first ]0
test tria]s) effective]y competed with the directed movements from the
interp0]ated S - S observationa] ]earning. Thus, in the stimu]us con-
text of the test chamber, Group TBR had a response that directIy
competed with the directed movements from the new (reversed) stimu]us—
reinforcer association, so that this group did not reverse on test
in Phase 5. However, Group RBT had no response in the context of
the test chamber that cou]d compete with the directed movements
dependent on the new (reversed) stimu]us—reinforcer association, so
that this group did reverse on test in Phase 3. After Group RBT had
a response estab]ished (from Phase 3) in the context of the test
chamber that cou]d compete with the directed movements from the new

(second reversa]) stimu]us-reinforcer association, this group a]so

]5
wou]d not reverse when tested in Phase 5. Even though the subjects
had ]earned the S — S reversa], as evidenced by reaching reversa]
discrimination criterion, subjects fai]ed to reverse on test on the
basis of S - S ]earning. The new S+ bar (after each reversa]) which
was more predictive of the reinforcer was not the bar which was
approached and contacted if the subject had received S - R ]earning
tria]s previous]y.

To summarize, subjects can acquire the response demonstrated
mechanica]]y in a mode]-absent situation on the basis of ]earning S - S
re]ationships (gaining information about environmenta] contingencies).
Furthermore, if prior S - R ]earning competes with directed movement
from new reversed S - S observationa] ]earning, the no reversa] wi]]
occur on test. If there is no prior S - R ]earning competing with
the directed movements from the reversed S - S ]earning, then reversa]
wi]] occur on test. Persistence of the S - S directed response was
determined by the absence of the occurrence of a competing response in
the stimu]us context of the observationa] apparatus (Denny & Ade]man,
]955).

The present experiments extended this ana]ysis 0f the processes
which under]ie the ]earning of responses by observation of stimu]us
sequences using the observationa] ]earning procedure. The first of
the three experiments was concerned with specifying the ]ocus of contro]
of the directed actions. One of the critica] questions raised by the
observationa] ]earning experiments was how important progressive]y
c]oser approach movements were for deve]opment 0f the fina] directed
response of barpressing and for the ]earning of the stimu]us-reinforcer

re]ation. As Wesse]]s (T974) pointed out, it was important that the

]6
directed movements emp]oyed as the index of stimu]us-stimu]us ]earning
not be conditioned in the observation phase. Any overt response condi-
tioned in the tunne] (such as directiona]]y differentiated orienting)
wou]d e]iminate the c]ear—cut separation of the S - S and the S - R
]earning, and wou]d bring into question which association was directing
responding on test tria]s. The first experiment contro]]ed for certain
incipient orienting responses in the tunne]. Experiment 2 assessed the
genera]izabi]ity of this phenomenon by testing for any sexua] dimorphism
in the ]earning and retention of an observationa] discrimination. A
number of other nonreproductive behaviors have been identified as being
sexua]]y dimorphic in rodents (see Beatty, ]979, for a review). The
]ast 0f the three experiments assessed the effects of certain pretraining
manipu]ations on the acquisition of directed responses. In the origina]
experiment, each subject typica]]y received five food-on]y (magazine
training) tria]s, though occasiona]]y a subject wou]d require more tria]s
before it was adequate]y magazine trained. The contribution of magazine
training to autoshaping has been demonstrated to be an important vari-
ab]e in the acquisition of directed responses (B1anchard & Honig, T976;
Downing & Neuringer, ]976; Engberg, Hansen, We]ker, & Thomas, ]972). In
Experiment 3, the number of US-onTy tria]s in pretaining was manipu]ated
to determine whether this variab]e cou]d be a source of variation among

subjects in performance on test tria]s.

EXPERIMENT T

In the design of the origina] set of experiments on mode]-absent
observationa] ]earning (Denny, Be]],& C]os, Note ]), subjects cou]d
make incipient orienting responses toward a particu]ar ]ocation (right
or ]eft) during the discrimination ]earning tria]s. Incipient
orientation toward a bar ]ocation, though not regu]ar]y observed, cou]d
have occurred whi]e the anima] was waiting for a tria] to begin. Thus,
there may have been S — R p]ace ]earning occurring whi]e the rat was
observing from the tunne]. Such ]earning cou]d have been responsib]e
during the testing phase for directing the anima] toward the S+ bar,
rather than the correct response during test having been mediated by
the S - S observationa] ]earning. Indeed, the resu]ts wou]d be trivia]
if a]] the subjects had ]earned was to “turn right” or "turn ]eft"
during the visua] discrimination phase and had not ]earned anything
about moving manipu]anda. Experiment ] was conducted to e]iminate
this possibi]ity as an exp]anati0n for the resu]ts from previous
experiments.

If rats were using S — R p]ace ]earning to direct them to an area
during test tria]s we wou]d expect poorer test tria] performance when
the bars are switched from side to side. On the other hand, making
the position cues irre]evant by switching the pane]s from side to side
cou]d have the resu]t of enhancing stimu]us contro] by the moving bars

and thereby faci]itate S - S ]earning and performance during test

17

]8
tria]s. Indeed evidence from Kohn and Dennis (]972) suggests that
the attention-getting va]ue of changing cue pane]s might yie]d faster
]earning of an observationa] discrimination task.

Furthermore, making a saTient dimension such as position irre]evant
cou]d, as Suther]and and Mackintosh (]97]) wou]d predict, produce an
overTearning reversa] effect - a faci]itation of reversa] performance
when a group is given overtraining tria]s prior to reversa]. In which
case, Group TBR that receives ]20 overtraining tria]s during Phase 3
wou]d be expected to show reversa] on Phase 5.
1m

Subjects. Twe]ve naive ma]e aTbino, Sprague-DaWTey, rats approxi-
mate]y ]50-200 days 0]d at the start of training were maintained at 90%
of ad ]ibitum weight (previous research showed that more high]y
deprived rats bit the tunne] door and visited the food tray excessive]y
and were thus poor subjects). Subjects were fed suppTementary food
in their home cages fo]]owing each experimenta] session to maintain
target weights. The rats were housed individua]]y under conditions
of a ]2-h ]ight/dark cyc]e, and a]] had free access to water in the
c0]ony throughout the experiment.

Apparatus. The experimenta] chamber for this study can be seen
in Figure ]. The funne]-shaped compartment containing the bars is
the test chamber. The hidden experimenter activated the bars from
behind the end pane] at a]] times during training. The tip of the
activated bar on the ]eft pane] in Figure ] has a tota] excursion of
8.5 cm and makes a sharp c]ick when it strikes the chamber f]00r. The
bar on the right pane] in Figure I, when activated, has an excursion

of 2.2 cm, turns on a dim ]ight above the bar and produces a faint

]9
c]ick from the microswitch. The cei]ing over the tunne] is opaque, the
one over the bars transparent. The cue pane]s cou]d be switched from
side to side.

Procedure. Ha]f the subjects in each group were trained to the
bar and ]ight (right stimu]us in Figure ]) as the positive stimu]us,
and ha]f were trained to the bar on]y (]eft stimu]us in Figure ]) as
the positive stimu]us. The first session inc]uded the three phases of
pre]iminary training: habituation, magazine training, and shaping of
the observing response. Rats were a]]owed to exp]ore both ha]ves of
the observation apparatus with the bars inoperab]e. At ]east ten 97
mg Noyes pe]]ets were avai]abTe in the food tray during exp]oration.
Once the rat ate these pe]]ets, five additiona] reward pe]]ets were
individua]]y de]ivered after de]ays of approximate]y 60 seconds whi]e
the rat was confined to the tunne] (magazine training). The observing
response was shaped by de]ivering further pe]]ets on]y after the rat
had ]eft the food tray and faced the c]osed tunne] door.

Fo]]owing pre]iminary training subjects were random]y assigned to
the two treatment groups and were shifted to the observation of the
contingencies phase. The procedure for this phase differed from
previous experiments on]y in so far as cue pane] position was concerned.
The pane]s that the manipu]anda were ]ocated on were switched from side
to side according to a computer generated hypergeometric distribution
of switches (p = .3). Because there was some attendant noise associated
with the changing of pane]s, phoney switches were schedu]ed a]so
(p = .32). This procedure contro]]ed for response a]ternation with
each switch. During the phoney switch the pane]s were removed from the
back wa]] of the observation apparatus and then rep]aced in the

origina] position. An opague pane] was p]aced in front of the tunne]

20

door during a]] switches. The ITI was a computer generated Poisson
distribution of time interva]s with a mean of ]0 seconds and a range
of ] to 60 seconds. This meant that wait-time cou]d not become a cue
for ]eaving the door and approaching the food tray. The ITI started
after the rat returned to the door from the food tray on S+ tria]s or
after the rat had waited ]0 sec at the door on S- tria]s. Extra visits
to the food tray during the ITI on]y increased the ITI if the rat fai]ed
to return to the door by the end of the ITI. Seven seconds were added
to a]] ITI va]ues, however, to a]]ow for time to make the pane] switches.

Testing began for subjects in Group TBR (Tested Before Reversed)
when they reached the discrimination criterion outTined in the
introduction. The cue positions were changed fo]]owing the series
used by Fe]]ows (]967). B]ocks 0f tria]s were chosen on test days so
that ha]f the time the tested pane] and the ]ast training tria] had the
8+ on the same side, and ha]f the time they were on opposite sides.
The opaque pane] was p]aced in front of the tunne] door prior to a]]
testing, and either a rea] or a phoney switch occurred. If the rat
pressed a bar it was removed from the test chamber and returned to its
home cage. Contacts to the ]eft bar (bar on]y stimu]us) were recorded
as barpresses if the ]ever was dispTaced from the "up” position
(.45 rad from horizonta]) to at ]east the horizonta] position. Contacts
to the right bar (bar and ]ight stimu]us) were recorded as barpresses
if the microswitch was tripped and the ]ight turned on. If the rat did
not press a bar within 5 minutes, it was returned to its home cage.
During testing, both the ]atency to barpress and the particu]ar bar

pressed were recorded.

2]

On each of three Succeeding days (test days 2, 3, 4) the rat was
p]aced in the tunne] and given 20 booster tria]s fo]]owed immediateTy
by a test as out]ined above. On test days 5, 6, 7, the rat was tested
both prior to and fo]]owing the b]ock 0f 20 tria]s. This gave a tota]
of ]0 tests over a 7 day period. Fo]]owing the 7 test days these
subjects received further observationa] training with the cues re—
versed. Upon reaching the same criterion as used for the origina]
discrimination the subjects in Group TBR were again given ]0 test
tria]s over 7 days.

The other treatment group, Group RBT, received identica] treat-
ment as the first group unti] the origina] discrimination criterion
was met. Upon reaching criterion, Group RBT was immediateTy reversed
and given more observationa] training tria]s. When Group RBT reached
reversa] discrimination criterion, the testing phase began as out]ined
above. Fo]]owing the 7 test days, these subjects received further
observationa] training with the cues reversed a second time. These
subjects comp]eted the experiment with a fina] testing phase.

Resu]ts and Discussion

 

Tab]e 3 summarizes the performance of subjects on discrimination
tria]s and test tria]s for the five experimenta] phases in
Experiment ]. Subjects were ab]e to ]earn the go/no-go discrimination
with switching cue pane]s in a mean of 240.8 tria]s (N = ]2). This
mean was somewhat greater than the mean number of tria]s to initia]
discrimination criterion as found by Denny, Be]], and C105 (Note I)
using stationary pane]s. Thus, S — S ]earning was not faci]itated
(stimu]us contro] by the moving bars was not enhanced) by switching

cue pane]s, at ]east with respect to number of tria]s to reach

22

discrimination criterion. This finding is inconsistent with Kohn and
Dennis (]972) who found that simp]y changing cue pane] ]ocation resu]ted
in a faci]itation of the observationa] ]earning task. Their findings,
though, were probab]y a resu]t of the changing stimuTation maintaining
the subject's attention, and in this manner providing for an enhanced
distinctiveness of cues. Whereas in the present experiment, presumab]y
attention was sufficient]y maintained by the presence of reinforcement
in the situation.

The switching of pane]s from side to side did, however, have a
significant effect on test tria] performance. As a group, TBR
subjects contacted the S+ bar on on]y 36.7% of the ]0 test tria]s.
Ha]f of the subjects did not contact either bar within 5 minutes after
test entry on 2 to 7 of the test tria]s during the first testing phase
(Phase 3). This can be compared to the TBR group without switching
cue pane]s in the Denny, Be]], and C]os study (Note ]) in which a]]
subjects contacted one or the other bar on a]] 10 test tria]s, and as
a group contacted the S+ bar on 86.7% of test tria]s (see Tab]e 2).
Consistent with the overa]] poor performance of Group TBR on test was
the poor performance on the initia] test tria]. The initia] test
tria] was critica] for assessing the S - S ]earning. Contact to the
bar on the initia] test wou]d presumab]y be directed by the S - S
observationa] ]earning. On subsequent test tria]s, however, conditioned
reinforcement (]ight, or c]ick of bars) or some other response
maintaining mechanism such as "marking“ (Lieberman, McIntosh, Thomas,
]979) cou]d have been directing responding a]ong with the S - S
]earning. On]y 3 out of 6 anima]s in the TBR group contacted the S+

bar on initia] test (compared to 6 out of 6 subjects without cue

Data from the five phases of Experiment 1.

23
Tab]e 3

Number of tria]s to observationa]

]earning criterion (inc]uding criteria] day) and percentage barpress response to
stimu]i over 10 test tria]s are ]isted

 

 

 

 

 

Group Subject Phase
1 2 3 4 5
TBR R 105 (8+) 232 - (8+) 0% (BL+) 200 (B-) 70%
(BL-) 30% (BL+) 30%
R 20] (8+) 20] - (8+) 60% (BL+) 198 (B-) 30%
(BL-) 40% (BL+) 70%
R 40] (8+) 288 - (8+) 10% (BL+) 317 (B-) 10%
(BL-) 90% (BL+) 90%
R 60] (BL+) 308 - (8-) 10% (8+) 280 (8+) 0%
(BL+) 70% (BL-) 100%
R 504 (BL+) 295 - (8-) 20% (8+) 228 (8+) 30%
(BL+) 50% (BL-) 70%
R 101 (BL+) ]88 - (B-) 70% (8+) 200 (8+) 50%
(BL+) 30% (BL-) 50%
Mean 252.0 - (+) 36.7% 237.2 (+) 45%
(-) 43.4% (-) 55%
5.0. 51.7 - - 50.2 -
RBT R 109 (BL+) 252 (8+) 360 (3+) 70% (BL+) 227 (8-) 80%
(BL-) 30% (BL+) 20%
R 160 (BL+) ]88 (8+) 307 (8+) 80% (BL+) 440 (B-) 100%
(BL-) 20% (BL+) 0%
R 130 (BL+) 243 (8+) 348 (8+) 90% (BL+) 320 (B-) 100%
(BL-) 10% (BL+) 0%
R ]50 (8+) 268 (BL+) 442 (8-) 20% (8+) 360 (8+) 10%
(BL+) 80% (BL-) 90%
R 106 (8+) 232 (BL+) 320 (B-) 0% (8+) 240 (8+) 0%
(BL+) 100% (BL-) 100%
R 500 (8+) 194 (BL+) 380 (B-) 20% (8+) 312 (8+) 30%
(BL+) 80% (BL-) 70%
Mean 229.5 359.5 (+) 83.3% 316.5 (+) 10%
(-) 16.6% (- n
S.D. 32.1 48.3 - 78.8 -
Note. The absence of a response on a test is indicated by a less than 100% when the

percentage response to the two stimu]i in that phase are summated.
training and test stimu]i in each phase for that subject.

Parentheses indicate
TBR = Tested Before Reversed;

RBT = Reversed Before Tested; 8 = Bar on]y stimu]us; BL = Bar and Light stimu]us.

24
pane]s changing). Thus, TBR subjects showed ]itt]e evidence of a
directed response from the S - S ]earning. As an additiona] measure,
]atency can be used to determine the directedness of the responses that
did occur. The median ]atency to contact was ]25 sec as compared with
a median ]atency of 49 sec for groups showing S - S directed responding
in ear]ier experiments with cue pane]s stationary. Informa]
observation of TBR anima]s on test showed that prior to contacting a
bar, time was spent exp]oring the chamber, rearing, or crouching in
front of the tunne] door. With on]y the moving bars and attendant
stimu]i for cues (without position as a cue), TBR subjects performed
poor]y on test on a variety of measures. Thus,it wou]d appear that
switching cue pane]s adverse1y affected test performance. This finding
1ends support to the notion that S - R p]ace ]earning may have been
directing subjects in previous experiments toward an area where the S+
bar was ]ocated.

However, in contrast to Group TBR, subjects reversed before
testing (RBT) performed on test much ]ike RBT subjects whose cue pane]s
were not switched. That is, subjects did reverse on test as a resu]t of
S - S ]earning in Phase 2. RBT subjects in Phase 3 pressed the reversed
(new S+) bar 83.3% of the time. However, these same subjects which
reversed once on the basis of S - S ]earning fai]ed to reverse on
test a second time on the basis of S - 5 after the intervening S - R
]earning (Phase 3). As can be seen in Tab]e 3, Group RBT in Phase 5
pressed the new S+ bar on]y 10% of the time. In other words, the
group continued to press the 01d S+ (new S-) 90% of the time. Five
out of six of these RBT anima]s on the initia] test tria] in Phase 3

pressed the bar most recent1y associated with the reinforcer, with

25
a]] subjects pressing one or the other bar on a]] subsequent test
tria]s. The median ]atency to barpress was 31.8 sec. Overa]], then,
RBT subjects were more directed than Group TBR in their responding.
Furthermore, this group showed directed responding without having S - R
p]ace ]earning to guide responding. So subjects can ]earn and perform
on test without position as a cue.

The greatest effect of switching the pane]s was on reversa]
performance. Group TBR received ]20 additiona] (overtraining) tria]s
across 6 test days (the booster tria]s) prior to reversa] training.
Group TBR ]earned the go/no-go reversa] in a mean of 237.2 tria]s as
compared with Group RBT that required a mean of 359.5 tria]s. Group
RBT was reversed without these extra tria]s, or overtraining. This
difference on tria]s to reversa] criterion was significant (Mann-
Whitney Two-Samp1e Test, p_< 0.025). In other words, there was a
significant positive 0ver]earning reversa] effect (ORE) when cue pane]s
were switched. In ear]ier studies without switched cue pane]s, no
significant ORE was found (Mann-Whitney Two-Samp1e Test, p_> 0.1).

The ORE observed in Group TBR, however, did not extend to bar-
pressing (test phase). After reversa] training, TBR subjects on test
(Phase 5) contacted the new S+ bar on on]y 45% of the test tria]s. As
in Phase 3 there was no consistent responding as a group to one bar
or the other.

The question remains why Group TBR in contrast to Group RBT
fai]ed to show consistent barpressing in either of the two test phases.
One tenab1e exp1anation fo]]ows from an S - R interpretation of the
data. Within this framework, ]earning depends upon consistentTy

e]iciting the to-be-1earned response in c]ose tempora] contiguity

26
with a particu]ar stimu]us situation a]ong with the minimization of
the e]icitation 0f a]ternative responses to the same or simi]ar
stimu]us situations (Denny & Ade]man, ]955). In this way, the response
that is being ]earned wins out over other possib1e responses. However,
if a]ternative responses are a]]owed to occur in a given situation
then those responses are the ones that are ]earned. Many variab]es
which he1p mediate ]earning are those that common1y minimize or
e]iminate competing responses. Habituation to the experimenta]
chamber with repeated exposure to the stimu]us situation, reduces the
nove]ty of the situation minimizing exp]oratory responses. It is
possib1e that habituation 0f exp]oration from repeated exposure to
the stimu]us situation accounts for the more consistent test tria]
performance of Group RBT as compared with Group TBR.

Group RBT with the changing cue pane]s took more tria]s to reach
the reversa] criterion and, therefore, was exposed to the experimenta]
situation for a ]ong time prior to ever being a]]owed into the test
chamber for testing. Such exposure cou]d have habituated any ex-
p]oratory responses that might compete with S — S directed responding
in the stimu]us situation of the test chamber. However, exp]oratory
responses may not have habituated by the time TBR reached discrimination
criterion (]ess exp05ure to stimu]us situation). Thus, when TBR subjects
were tested, these a]ternative responses were in direct competition
with the S - S directed barpress responses, resu]ting in inconsistent
performance. For some of the subjects, the a]ternative responses
(which inc]uded rearing, nosing about, and crouching) won out, and
no barpress occurred on from 2 to 7 of the test tria]s. For other
subjects, this response competition was indicated by ]ong ]atencies

when the responses did occur.

27

Group TBR continued to perform inconsistentTy on retest even
though by Phase 5 it had approximate]y as much exposure to the stimu]us
situation as Group RBT had had when it was first tested. In other
words, Group TBR in the second test phase shou1d have had as much
habituation of exp]oration as RBT had in the first test phase and
might be expected to barpress consistent1y. However, according to our
S - R position, since these a]ternative responses in Group TBR occurred
in the test chamber in the first test phase, these were the responses
that were ]earned. These a]ternative responses ]earned in the first
test phase were sti]] in direct competition with the new S — S
directed responding and therefore prevented consistent barpressing in

Phase 5.

EXPERIMENT 2

There has been a ]ongstanding tradition in anima] psycho]ogy
(especia11y rodent research) to conduct experiments on ma]e subjects,
perhaps because of the ear]y discovery of estrus ]inked changes in
activity (S10naker, 1925). Fema1es rats are on a four-day rhythm in
gross bodin activity associated with the period of oestrum. Therefore,
to avoid introducing unwanted and extraneous variabi1ity that might
obscure the phenomena of interest, the wide]y practiced research
strategy of exc]uding fema]es has been adopted.

Despite this ]ong tradition of designing experiments in a manner
that prec1udes discovery of sex differences in behavior, many behaviors
have been described as being sexua]]y dimorphic in rodents.
Differences between ma]e and fema]e rats have been found to be the
ru1e rather than the exception in aversive1y motivated ]earning
(Van Oyen, Van de P01], & de Bruin, 1979). For exampTe, superior
performance by fema]es has been found in aversive contro] procedures
in which the presentation of an aversion stimu]us is avoided or
terminated by a response. Whi]e fema]es are superior to ma]es in
active avoidance, they exhibit inferior performance on passive
avoidance tasks (Beatty, Gregoire, & Parmiter, 1973). The ]earning
and retention of certain kinds of mazes have a]so been shown to be
sexua]]y dimorphic in rats. In genera], ma]es are superior (Barrett

& Ray, 1970), with comp]ex mazes being most sensitive to sex differences.

28

29

Genera] sex differences in emotiona]ity(Gray, 1979) and the
activationa] and organizationa] effects of gonada] hormones have been
offered as being responsib]e for the manifestation of sex differences
in ]earning tasks (see Beatty, ]979, for a review). Specifica11y,
testosterone proprionate has been imp1icated in inducing three genera]
c]asses of change: (a) an increase in persistence of response to a
particu]ar type of stimu]us, (b) an increase in persistence of
response to stimu]i in a particu]ar p]ace, and (c) an increase in
resistance to distraction (especia11y with respect to irre]evant
stimu]i) coup1ed with an increased abiTity to sustain attention on a
particu]ar ]oca1ized stimu]us (see Andrew, 1976, for a review, though
he on]y discusses chicks and mice). Sexua] dimorphism on task wou]d
be expected, according to Andrew, to the extent that persistence and
attentiona] processes affect the success or fai]ure on a task. As
both processes are critica] to success in the observationa] ]earning
situation, sex differences wou]d be expected.
32cm

Subjects. Eighteen naive fema]e hooded and a1bino rats approxi-
mate]y 150-200 days o]d at the start of training served as subjects.
Rats were housed under the same conditions as subjects in Experiment ].
Deprivation ]eve1s were a]so the same as for subjects in Experiment 1.

Apparatus. The experimenta] chamber used in Experiment 1 was
used in this experiment, with one modification - cue pane]s were
stationary. The bar and ]ight cue pane] remained on the right side
of the front wa]], whi]e the bar on]y pane] remained on the ]eft side

(see Figure 1).

30

Procedure. The eighteen subjects were random]y assigned to three
treatment groups. Two of the groups received the same treatment as
ma]es received in Group TBR and Group RBT in the origina] design of the
experiment as out]ined in Tab]e 1. Performance during observationa]
training tria]s (tria]s to discrimination criterion), ]atency to
barpress on test, and the particu]ar bar pressed were recorded. The
third group, Group C—TBR, received the same treatment as Group TBR
except that this group did not receive the b]ock of 20 (booster)
tria]s on each of the test days. This group was added to contro1 for
the possibi]ity that the 120 overtraining tria]s (over the course of
the 7-day test period) affected the course of discrimination reversa]
training or performance on test tria]s fo]]owing reversa] training.
E1iminating the extra 120 tria]s in Group C-TBR better equated the
treatment received by the two groups (RBT and C-TBR) prior to reversa].
The one critica] difference that remained was testing prior to or
fo]]owing reversa] of S - S ]earning.

Resu]ts and Discussion

 

Tab]e 4 summarizes the performance of subjects on discrimination
tria]s and test tria]s for the five experimenta] phases in
Experiment 2. Fema1es ]earned the go/no-go discrimination in a mean
of 227.6 tria]s (N = 18). Ma1es from Denny, Be]], and C]os (Note 1)
]earned the same discrimination mean of 192.2 tria]s. A1though
compared across experiments, the number of tria]s to first
discrimination criterion was not significant]y different for fema]es
and ma]es (Mann-Whitney Two-Samp1e Test, p_> 0.1). Even so, fema]e
performance on ear]y tria]s differed marked1y from that of ma]es.

A1though not systematica11y recorded, informa] observation indicated

31

Tab]e 4

Data from five phases of Experiment 2. Number of trials to discrimination criterion
(inc]uding criteria] day) and percentage barpress response to stimu]i over 10 test
tria]s are ]isted

 

 

 

 

 

Group Subject Phase
1 2 3 4 5
TBR R 702 (8+) 274 - (8+) 30% (BL*) 610 (B-) 30%
(BL-) 70% (BL+) 701
R 706 (8+) 180 - (3+) 01 (BL+) 320 (B-) 20%
(BL-) 1001 (3L+) 80%
R 803 (BL+) 184 - (B-) 301 (8+) 314 (8+) 01
(BL+) 60% (BL-) 100%
R 705 (3L+) 391 - (3-) 20% (3+) 430 (3+) 30%
(BL+) 80% (BL-) 70%
R 704 (BL+) 220 - (B-) 60% (8+) 380 (8+) 60%
(BL+) 40% (BL-) 40%
Mean 249.8 - (+) 42% 410.8 (*) 521
(~) 56% (-) 481
S.D. 87.5 - - 121.1 -
RBT R 707 (BL+) 180 (8+) 272 (8+) 701 (BL+) 280 (B-) 60%
(BL-) 30% (BL+) 40%
R 701 (BL+) 180 (8+) 477 (0+) 901 (BL+) 520 (B-) 1001
(BL-) 10% (BL+) 0:
R 703 (8+) 296 (BL+) 660 (8-) 40% (BL+) 365 (8+) 50%
(BL+) 60% (BL-) 50%
R 708 (8+) 268 (BL+) 267 (B-) 0% (B4) 300 (8+) 0%
(BL+) 100% (BL+) 100%
R 709 (3+) 183 (BL+) 300 (B-) 10% (8+) 347 (9+) 0%
(BL+) 90% (BL-) 1001
Mean 221.4 395.2 (+) 822 362.4 (+) 18%
(-) 181 (-) 82%
5.0 56.2 171.4 - 94.6 -
C-TBR R 804 (8+) 264 - (3+) 20% (BL+) 579 (B-) 10%
(BL-) 80% (BL+) 90%
R 306 (8+) 390 - (8+) 50% (BL+) 584 (B-) 50%
(BL-) 50% (BL+) 50%
R 212 (8*) 320 - (8+) 50% (BL+) 560 (B-) 0%
(BL-) 50% (BL+) 100%
R 017 (3+) 220 - (8*) 60% (BL+) 560 (B-) 70%
(BL-) 40% (BL+) 30%
R 801 (BL+) 260 - (B-) 0% (8+) 520 (8+) 0%
(BL+) 100% (BL-) 1001
R 805 (BL+) 306 - (B—) 30% (8+) 470 (8+) 80%
(BL+) 70% (BL-) 20%
R 211 (BL+) 200 - (B-) 10% (8+) 480 (B+) 10%
(BL+) 90% (BL-) 90%
R 018 (BL+) 200 - (B~) 40% (8+) 400 (8+) 100%
(BL+) 10% (BL-) 01
Mean 270 - (+) 56.3% 519.1 (+) 63.7%
(-) 37.5 H 36.3%
5.0. 66.2 - - 64.7 -
Note. The absence of response on a test tria] is indicated by a 1ess than 1001 when the

 

percentage response to the two stimu]i in that phase are sunnute arentheses indicate
training and test stimu]i in each phase for that subject. TBR . Tested Before Reversed;

RBT = Reversed Before Tested; C-TBR . Contr01-Tested Before Reversed; B ' Bar on]y Stimuius;
8L = Bar and Light stimu]us

32
that the fema]es' performance on ear]y tria]s was characterized by ]ong
periods of crouching at the back of the tunne], much defecation and
urination, and a re1uctance to approach and eat from the food tray or
to approach the front of the tunne]. Some fema]es wou]d take an hour
of habituation before they wou]d eat from the food tray. When food
pe]]ets were dropped through the tube into the tray the attendant noise
wou]d often e]icit freezing. Long pretraining periods were necessitated
by the subjects' behavior, inc]uding more magazine training tria]s and
extended training to shape the observing response. The pretraining
phase took on the order of severa1 days, a]though the number of pe]]ets
de]ivered during this extended pretraining phase was not recorded.
Ma1es, on the other hand, typica]]y ]earned the observing response
(]earned to shutt1e back to the tunne] door in 5 tria]s) and readi1y
ate from the food tray.

Group and sex differences were found on test. As a group, TBR
fema]es contacted the S+ bar on on]y 42% of the first 10 test tria]s.
Performance within the fema]e TBR group was marked by variabi1ity,
with subjects pressing the S+ bar on from 0 to 8 of the 10 test tria]s.
TBR ma]es contacted the S+ bar on 86.7% of the test tria]s, with a
range from 7 to 10 of the 10 test tria]s. Consistent with the TBR
fema]es' overa]] poor performance on test, was the poor performance on
the initia] test tria]. On]y 2 out of 5 fema]es in the group pressed the
S+ bar on the initia] test (a]] TBR ma]es from Denny, Be]], and C]os,
Note 1, pressed the S+ bar on initia] test). Thus, the TBR fema]es on
test showed ]itt]e evidence of an S+ directed response on the basis of

S - S ]earning.

33

Whi]e TBR fema]es did not as a group press the S+ bar, they
did, however, press one bar or the other fair1y consistent1y, with
on]y one fema]e not pressing a bar within 5 minutes of test entry.
Indeed, a]though responses were as often directed at S- as at S+,
fema]es responded quick1y upon test entry, with a median ]atency of
33 sec (the median ]atency for TBR ma]es was 49 sec).

TBR fema]es reached reversa] criterion in a mean of 410.8 tria]s
(TBR ma]es reached criterion in 192.2 tria]s, Tab]e 2). A1though it
took fema]es many more tria]s to ]earn the go/no-go reversa], the
fema]es on test performed ]ike the ma]es in that they did not reverse.
As a group, TBR fema]es pressed the S+ bar on]y 52% of the time,
however, those fema]es that c0nsistent1y pressed a bar on the first
set of test tria]s, continued to do so on reversa] whether that
stimu]us was the 5+ or the S-. Thus, again, S — R ]earning (consistent
responding to a particu]ar bar) directed responding on test. TBR
fema]es, ]ike ma]es, fai]ed to reverse on test on the basis of S - S
]earning. Un1ike ma]es, though, TBR fema]es as a group c0nsistent1y
fai]ed to approach the S+ bar in Phase 3 and 5.

In contrast to the performance of TBR fema]es, which was marked1y
discrepant from ma]e TBR performance on most measures (tria]s to
reversa] criterion, initia] test tria] performance, and mean test
performance), Group RBT fema]es responded simiar1y to ma]es on at
]east some measures. RBT fema]es' mean percent response to the new S+
(01d S-) in Phase 3 was 82% (range from 60 to 100 percent), as compared
with a mean of 92% for RBT ma]es (range from 90 to 100 percent). Ma1es
were somewhat superior to fema]es on test performance. C1ear1y, though

both RBT ma]es and fema]es reversed on the basis of S - S ]earning.

34
Moreover, RBT fema]es, ]ike RBT ma]es that reached reversa] criterion
on the go/no-go discrimination a second time (Phase 4), fai]ed to
reverse on test, responding instead to the 01d S+ on 82% of the test
tria]s. After intervening S - R tria]s both sexes fai]ed to reverse
on the basis of S - S ]earning in the tunne].

Thus, an S - R interpretation of the reversa] data from the
modeT-absent observationa] ]earning situation fits for both sexes. As
described ear]ier, the persistence of a directed response after
repeated testing is presumab]y due to the absence of a competing
response in the stimu]us context of the test chamber. There remains,
however, another p]ausib1e exp1anation for the difference between
groups RBT and TBR 0n reversa] performance. On]y group TBR received
120 overtraining (booster) tria]s prior to reversa]. It is possib1e
that the overtraining affected reversa] performance both in terms of
tria]s to reversa] criterion and test tria] performance. Group TBR
]earned the go/no-go reversa] in a mean of 410.8 tria]s, and Group
RBT ]earned the reversa] in a mean of 395.2 tria]s. This sma1]
difference was not significant (Mann-Whitney Two-Samp1e Test,

p > 0.1). The overtrained group (TBR) did not reverse on test, whereas
Group RBT did reverse on initia] test. A more direct test of the

effect of 120 overtraining tria]s came from comparing resu]ts of

Group C-TBR (without 120 booster tria]s) with TBR. Like the TBR

group, fema]es in C-TBR fai]ed to show consistent S+ directed responding
on test tria]s (mean of 56.3% correct in Phase 3 and 63.7% in Phase 5).
No significant differences were found between C-TBR and TBR in either
the number of tria]s to reversa] criterion on the go/no-go discrimination

or in the test tria] performance fo]]owing reversa] (Mann-Whitney

35
Two-SampTe Test, p_> 0.1). Nothing significant about reversa]
performance is accounted for by the 120 overtraining tria]s.

A1though extended training (via the go/no—go reversa]) improved RBT
fema]e performance on test, the fact remains that sex differences in
S+ directed performance were c1ear1y found across groups. Ma1es show
a more persistent response attachment to a particu]ar stimu]us in a
particu]ar p]ace on the basis of S - S observationa] (cognitive)
]earning. This finding is consistent with work on search strategy
with chicks and mice (Andrew, 1972, 1975; Andrew & Rogers, I972)- In
these studies testosterone has been shown to increase persistence in
search for both a particu]ar type of food and in a particu]ar p]ace.
Such persistence inc]udes decreased distractibi1ity coup1ed with an
increased abi1ity to sustain attention on a particu]ar ]oca1ized
stimu]us. A wide variety of different experiments ]ead to the con-
c]usion that with the manipu]ation of testosterone, the centra]
specifications used in recognition of a particu]ar stimu]us are 1ike1y
to remain more persistent1y in use once activated. Comparab1e effects
of androgens on cognitive abi1ities have been reported in humans, so
that mechanisms of widespread importance in higher vertebrates may
be invo]ved. For instance, there is reason to suspect that gonada]
steroids may inf1uence certain cognitive ski11s - ma]es perform better
than fema]es on spatia1 measures and these sex differences do not
genera11y emerge unti] after puberty (Harris, L.J., 1978).

The resu]ts of Experiment 1 (position cues contro]]ed) and
Experiment 2 with fema]es were simi]ar in that Group TBR subjects
fai]ed to show 5+ directed responding. This finding suggests an

a]ternative exp1anation. It cou]d be that exp]oratory responses in

36
the test chamber had not habituated by the time TBR and C-TBR fema]es
reached discrimination criterion (1ess exposure to stimu]us situation)
so that these a]ternative responses were in direct competition with
the S - S directed barpress response. The inconsistent responding
observed on test in Groups TBR and C-TBR is in keeping with this
interpretation. In further support of this interpretation, Group RBT
fema]es given extended exposure to the situation prior to test
responded to the 5+ on 82% of the test tria]s. This extended exposure
(from a 1engthy reversa]) meant that subjects had brief but dai]y
access to the test chamber when p]aced in and removed from the
apparatus. This exposure cou]d have habituated any exp]oratory
responses prior to test. Thus responses in direct competition with
barpressing were e]iminated.

However, not a]] the data fit this interpretation. The fema]es
did not obvious1y exp]ore the test chamber prior to barpressing. The
short ]atencies to barpress on test in TBR and C-TBR subjects indicate
a ”directedness“ to their responses and a ]ack of competition from
other responses. A]so on]y 2 fema]es out of 13 tested before reversa]
(TBR and C—TBR combined) fai]ed to press a bar, as opposed to ha]f the
TBR subjects in Experiment 1.

There is yet another possib1e exp1anati0n for the sex difference
observed on test. The fema]es required a ]onger period of pretraining—
inc1uding more magazine training tria]s and extended training to
shape the observing response. If during this extended pretraining
a]] the stimu]i in the chamber were effective or re1evant cues (a]]
predictors of reinforcement), then such training cou]d retard

subsequent discrimination training where the brunt of stimu]us contro]

37
must be on one e]ement. B]ocking by the background cues cou]d occur.
Indeed, a number of pretraining manipu]ations have been shown to inter-
fere with the acquisition of a directed response (Downing & Neuringer,
1976; Engberg, Hansen, We]ker, & Thomas, 1972; Tomie, Murphy, Fath, &
Jackson, 1980; Wasserman, 1972). As there is no way of testing the
tenabi1ity of this exp1anation from the data gathered in Experiment 2,
Experiment 3 was designed to systematica11y manipu]ate amount of

pretraining.

EXPERIMENT 3

The amount of magazine training during pretraining has been demon-
strated to be an important variab]e in the acquisition of a directed
response (Downing et a]., 1976; Engberg et a]., 1972). Magazine
training exposes the subject to unpredictab1e presentations of the US,
and an excessive amount of such training interferes with the acquisition
of directed responses.

Wasserman (1972), whi]e examining the inf1uence of cue redundancy
on autoshaping, ran a no-cue pretraining group that received unsigna11ed
food presentations during the first phase of the experiment (10 days, 400
tria]s). Then the group received key]ight p1us c]icker tria]s fo]]owed
by food. In this test phase, the no-cue pretraining birds pecked infre-
quent1y and autoshaping was great1y retarded compared to c0ntr01 groups.
The birds typica]]y paced back and forth in front of the response pane].
Simi1ar findings were reported by Engberg et a]. (1972). They exposed
their birds to many more grain presentations than Wasserman (approximate]y
900 grain presentations, on a procedure ]ike Wasserman's no-cue group).
The resu]t of their manipu]ation was a significant retardation of
autoshaping.

Downing et a1. (1976) systematica11y varied the number of food-
onTy presentations instead of emp]oying on]y one va]ue as was done in
the previous1y cited studies. They were attempting to determine the

form of the function re]ating number of pre]iminary food-on]y tria]s

38

39
to how rapid1y pecking wou]d deve]op under autoshaping contingencies.
Four groups of chickens received either 1, 10, 100, or 1000 presenta-
tions of grain in a hopper. The number of key]ight—grain pairings before
a bird first pecked the ]ighted key was found to be a U-shaped function
of the number of prior food-on]y presentations, with pecks occurring
significant]y sooner after 100 food-on]y va]ues. This resu]t seems to
exp1ain why Mackintosh (1973) found no retardation effect of pre-
training contrary to Engberg et a]. (1972) and Wasserman (1972).
Mackintosh (1973) gave approximate]y 200 magazine presentations to
his food-on]y birds, which, according to the function from Downing et a].
(1976) sh0u1d not have retarded acquisition of autoshaping (a]though
this comparison is across species).

Investigators have most often appea1ed to cognitive factors such
as “]earned 1aziness” (Engberg et a]., 1972) in order to account for
the de1eterious effects of such pretraining on directed responses.
Engberg interpreted ”]earned 1aziness" as a para11e1 phenomenon to
“]earned he1p1essness" in an appetitive situation. The conception was
based on the premise that the retarded acquisition of autoshaping was
a genera] transfeF of training effect - a by—product of associative
impairment from pretraining. That is, during pretraining the subject
]earns that the US is unpredictab1e. This ]earning transfers to the
autoshaping situation, where it proactive1y interferes with the
directed response.

Tomie (1976) has offered an a]ternative interpretation of these
retardation effects based on the ”b]ocking“ phenomenon reported by
Kamin (1969). Tomie suggests that pretraining with unpredictab1e US

presentations may condition the static contextua] stimu]i present in

40

the conditioning environment during pretraining. This excitatory
context is then compounded with the i11uminated key CS during auto—
shaping, where it exerts a b]ocking inf1uence that resu]ts in the
retardation of acquisition. Empirica1 support for the context-b]ocking
interpretation comes from experiments uti1izing context-shift designs.
The resu]ts of Tomie et a1. (1980) casts some doubt on nonspecific
transfer interpretations by demonstrating that retardation is context
specific; gross changes in background contextua] stimu]i prior to
initiation of acquisition tests comp]ete1y a11eviate de1eterious effects
of pretraining.

Whi]e the retardation of directed responses has been we]] documented,
inc]uding the function describing the number of tria]s necessary to
effect a retardation, work has been confined to studies with pigeons
and chickens. There is no reason to expect that the function that
determines the number of food-on]y tria]s needed for retardation of
keypecking with pigeons and chickens wou]d h01d for barpressing with
rats. Additiona11y, there is no reason to expect that a function de-
rived from standard autoshaping studies wou]d app1y to the b]ocked
autoshaping design emp]oyed here. Therefore, in order to answer the
question of whether the extra pretraining given fema]es in Experiment 2
actua]]y retarded acquisition of the directed response, a new function
must be determined for rats in the present observationa] ]earning
situation.

Me_tho_d
Subject. Sixteen naive ma]e a1bino rats approximate]y 150-200

days 01d at the start of training served as subjects. Rats were

41
housed under the same conditions as subjects in Experiment 1.
Deprivation 1eve1s were a]so the same as for subjects in Experiment 1.

Apparatus. The experimenta] chamber used in Experiment 2 was
used in this experiment.

Procedure. The sixteen subjects were random]y assigned to four
treatment groups. Ha]f of the subjects in each group were trained to
the right stimu]us (bar and ]ight) as the positive stimu]us, and ha]f
were trained to the ]eft stimu]us (bar on]y) as the positive stimu]us.
During the first session, rats were a]]owed to exp]ore both ha]ves of
the observation apparatus with the bars inoperab]e. At ]east ten 97
mg Noyes pe]]ets were avai]ab1e in the food tray during exp]oration.
Next 1, 5, 35, or 100 food-on]y tria]s were schedu]ed according to a
computer generated Poisson distribution of time interva]s between
tria]s. The distribution had a mean of 10 seconds and a range of ] to
60 seconds, i.e., food was presented independent1y of behavior on the
average of six times per minute in the tunne] with the door c]osed. The
number of food—on]y tria]s given to the treatment groups was determined
from pi10t data which indicated that 100 tria]s did retard the acqui-
sition of the directed response. The first session with the food-on]y
presentations ended after the required number of tria]s was presented.

The observationa] phase of training began in the next session and
continued as in previous experiments unti] the discrimination criterion
was met. A]] anima]s were tested fo]]owing observationa] training. No
overtraining 0r booster tria]s were given on test days - that is, a]]
anima]s were in Group C—TBR. Subjects were not reversed, however, as

the question of interest was the affect of pretraining manipu]ations

42
on initia] acquisition and test. The number of tria]s to criterion
and performance on test tria]s were the major dependent variab]es.

Resu]ts and Discussion

 

Tab]e 5 summarizes the performance of subjects on discrimination
tria]s and test tria]s in Experiment 3. A Kruska1-Wa11is one way
ana1ysis of variance by ranks was conducted on the dependent measure,
tria]s to criterion, for the four groups (Siege1, 1956). Resu]ts con-
firmed that amount of magazine training (at ]east within the range of
va]ues measured) did not significant]y affect the number of tria]s to
criterion in the go/no-go discrimination: H (3) = 1.93566, p_> 0.25.

As can be seen in Figure 2, performance on the CS+ and CS- tria]s
differed as a function of the number of pretraining tria]s. A11 sub-
jects in Groups 1, 5, and 35 were responding re1iab1y (3_70% correct)

on CS+ tria]s within 80 tria]s, as compared with Group 100 with on]y

one subject responding re1iab1y within 80 tria]s. C1ear1y, conditioning
to the CS+ is retarded during the observationa] phase, and the magnitude
of this retardation is direct1y re]ated to the number of US-on1y tria]s.
Such data are consistent with a b]ocking interpretation. However,
evidence of b]ocking is camouf1aged when the overa]] discrimination
criterion is used for ana]ysis because of the overa]] ]ack of responding
to either stimu]us in Group 100.

Figure 3 shows that the number of pretraining tria]s was a signi-
ficant factor in test tria] performance. As a resu]t of pretraining
tria]s 1-35, performance increases monotonica11y, as shown by a signi-
ficant positive corre1ation between the 1ogarithm of the number of
pretraining tria]s and the number of correct responses (r = 0.48,

d.f. = 10, p = 0.05, one-sided). One hundred tria]s produces a

43

Tab]e 5

Performance of Subjects Grouped by Number of Pretraining Tria]s

 

Group Subject Tria]s to Criterion Percent Response

(5+ direction) to S+

1 - US On]y A-108 220 (R) 80%
A-203 315 (R) 100%

A-503 380 (L) 90%

A-305 235 (L) 0%

Mean 287.5 67.5%

5 - US On]y A-604 268 (R) 100%
A-306 415 (R) 90%

A-603 227 (L) 90%

A-605 255 (L) 70%

Mean 291.3 87.5%

35 - US On]y A-202 314 (R) 90%
A-602 325 (R) 80%

A-60] 260 (L) 100%

A-304 320 (L) 90%

Mean 304.8 90%

100 - US On]y A-107 180 (R) 60%
A-10] 160 (R) 10%

A-502 282 (L) 0%

A-501 320 (L) 10%

Mean 235.5 20%

 

 

 

   
 

44

 

 

  

 

CS+

 

 

 

 

100 US only

 

149

 

111111111111
123456789101112

100-
901— ———————————————————
cs-
80— “:1,
70-
34) 60-
=4
2 5015" '
2 46
U) _ 1USonly
1.11 30—
1:
U
m 10.. 10
E 01111111111119114 0
0 123456789101112
U
— cs+
E 100 100
3 9° """"""" . 9°
[1: n:
o: 80— 80
IE. 70* 70
Z _.
< 6° 1n=41 so
I“ so» 50
2
4°” 40
30-
35USonly 3°
20- 20
101' 10
oJlllllflmTJlLl 0
123456789101112
BLOCKS OF 40 TRIALS
Figure 2.

Mean percent responses on CS+ and CS- tria]s for each of
the four groups in Experiment 3. The p_represents the
number of subjects used in ca]cu]ating the mean percent
correct for that b]ock. The arrows indicate when subjects
reached overa]] criterion. Subjects reaching criterion
within a 40 tria] b]ock were inc]uded in ca]cu]ating the

mean for that b1ock.

45

 

 

 

100-

+01

0 90

l-

Lu 80

co

(2) 70

0.

f3 60

1:

0.1 50

‘3

)— 40

z

{‘3 30

a:

E 20

z

< 10—

DJ

E 01 1 1 #1

1 5 35 100

NUMBER OF PRETRAINING TRIALS

Figure 3. Mean percent correct response on test as a function of
the number of pretraining food-on]y tria]s. Note
abscissa is p]otted on a 1ogarithmic sca]e.

46
significant retarding effect when compared with 35 pretraining tria]s
(Mann-Whitney Two-Samp1e Test, p_< 0.01).

From the resu]ts of Experiment 3, it seems doubtfu] that the
differences found on test performance in Experiment 2 can be exp1ained
by "b]ocking" from contextua] cues since even the most difficu1t fe-
ma]es received fewer than 35 food-on]y tria]s. In fact, from the data
in Experiment 3, fema]es requiring more than 5 pretraining tria]s, up
to 35 tria]s, might have actua]]y shown increasing]y better performance

on test.

LIST OF REFERENCES

1.

Reference Note

Denny, M.R., 8e11, R.C., & C]os, C.F. Observationa], two-choice
]earning in the rat. Unpub1ished manuscript, 1980. (Avai1ab1e

from the first author at Michigan State University.)

47

LIST OF REFERENCES

Andrew, R.J. Recognition processes and behavior, with specia]
reference to effects of testosterone on persistence. In 0.5.

Lehrman, R.A. Hinde, & E. Shaw (Eds.), Advances in the study of

 

behavior (V01. 4). New York and London: Academic Press, 1972.
Andrew, R.J. Effects of testosterone on the behavior of the domestic
chick 11. Effects present in both sexes. Anima] Behavior, 1975,

2_3, 156-168.

 

Andrew, R.J. Attentiona] processes and anima] behavior. In P.P.G.
Batesch & R.A. Hinde (Eds.), Growing points in etho1ogy. London:
Cambridge University Press, 1976.

Andrew, R.J., & Rogers, L. Testosterone search behavior and
persistence. Nature (London), 1972, 237, 343-346.

Barrett, R.J., & Ray, 0.S. Behavior in the open fie1d. Lash1ey III
maze, shuttTe-box, and Sidman avoidance as a function of strain,
sex, and age. Deve10pmenta1 Psycho1ogy, 1970, 3, 73-77.

Beatty, W.W. Gonada] hormones and sex differences in nonreproductive
behaviors in rodents: 0rganizationa] and activationa] inf1uences.

Hormones and Behavior, 1979, 12, 112-163.

 

Beatty, W.W., Gregoire, K.C., & Parmiter, L.L. Sex differences in

retention of passive avoidance behavior in rats. Bu11etin of

 

Psychonomic Society, 1973, 2, 99-100.

 

48

49
Berry, C.S. The imitative tendency of white rats. J0urna1 of Comparative

Neuro]ogy and Psycho1ogy, 1906, lg, 333-361.

 

Berry, C.S. An experimenta] study of imitation in rats. Journa] of

 

Comparative Neuro]ogy and Psycho]ogy, 1908, 12, 1-25.
Bindra, D. How adaptive behavior is produced: A perceptuaT—motivationa1

a]ternative to response-reinforcement. Behaviora] and Brain

 

Sciences, 1978, 1, 41-91.

B1anchard, R., & Honig, W.K. Surprise va]ue of food determines its

effectiveness as a reinforcer. Journa] of Experimenta] Psycho1ogy:

 

Anima] Behavior Processes, 1976, 2, 67-74.

 

8011es, R.C. Reinforcement, expectancy and ]earning. Psycho]ogica1

 

Review, 1972, 22, 394-409.
Browne, M.P. The ro]e of primary reinforcement and overt movements in

autoshaping in the pigeon. Anima] Learning and Behavior, 1976,

 

3, 287-292.
Corson, J.A. Observationa] ]earning of a ]ever pressing response.

Psychonomic Science, 1967, Z, 197-198.

 

Davey, G.C., Oak]ey, 0., & C]e]and, G.G. Autoshaping in the rat:

Effects of omission on the form of the response. Journa] of the

 

Experimenta] Ana1ysis of Behavior, 198], 2g, 75-91.

 

De] Russo, J.E. Observationa] ]earning in hooded rats. Psychonomic

 

Science, 1971, 25, 37-38.
0e] Russo, J.E. Observationa] ]earning in discriminative avoidance

in hooded rats. Anima] Learning and Behavior, 1975, 2, 76-80.

 

Denny, M.R. A ]earning mode]. In W.C. Corning, & S.C. Rather (Eds.),

The chemist:y_of ]earning. New York: P1enum Press, 1967.

 

50
Denny, M.R. A theory of experimenta] extinction and its re]ation to a

genera] theory. In H.H. Kend1er, & J.T. Spence (Eds.), Essays in
neobehaviorism: A memoria] vo1ume to Kenneth S. Spence. New York:
Academic Press, 1971.

Denny, M.R., & Ade]man, H.M. E1icitation theory: I. An ana]ysis of

two typica] ]earning situations. Psycho]ogica1 Review, 1955, g2,

 

290-296.
Dodwe]], P.C., & Bessant, D.E. Learning without swimming in a water

maze. Journa] of Comparative and Physi01ogica1 Psycho]ogy, 1960,

 

52, 422-425.
Downing, K., & Neuringer, A. Autoshaping as a function of prior food

presentations. Journa] of the Experimenta] Ana1y§js of Behavior,

 

1976, 2Q, 463-469.

Dubanoski, R.A., & Parton, D.A. Effect of the presence of a human
mode] on imitative behavior in chi1dren. Deve10pmenta1 Psycho]ogy,
1971, 3, 489-497.

Engberg, L.A., Hansen, G., We]ker, R.L., & Thomas, D.R. Acquisition of
keypecking via autoshaping as a function of prior experience:
“Learned 1aziness"?, Science, 1972, 122, 1002-1004.

Estes, W.K. New perspectives on same 01d issues in association theory.
In N.J. Mackintosh, & W.K. Honig (Eds.), Fundamenta1 issues in

associative ]earning. Ha1ifax, N.S.: Da1housie University Press,

 

1969.

Estes, W.K. Reinforcement in human behavior. American Scientist, 1972,
29, 723-729.

Fe]]ows, B.J. Chance stimu]us sequences for discrimination tasks.

Psycho]ogica1 Bu11etin, 1967, p2, 87-92.

 

51
Gray, J.A. Sex differences in the emotiona] behavior of 1ab0ratory
rodents: Comment. British Journa] of Psycho]ogy, ]979, 22,
35-36.
Groesbeck, R.W., & Duerfe]dt, P.H. Some re1evant variab]es in

observationa] ]earning of the rat. Psychonomic Science, 1971,

 

22, 41-43.

Haggerty, M.E. Imitation in monkeys. Century Magazine, 1909, 22,

 

81-95.
Harris, L.J. Sex differences in spatia1 abi1ity: Possib1e environmenta],
genetic, and neuro]ogic factors. In M. Kinsbourne (Ed.),

Asymmetrica] function of the brain. Cambridge: Cambridge University

 

Press, 1978.
Hayes, K.J., & Hayes, C. Imitation in a home raised chimpanzee. Journa]

of Comparative and Physio]ogica1 Psycho]ogy, 1952, 32, 450-459.

 

Hearst, E. Stimu]us re]ationships and feature se]ection in ]earning and
behavior. In S.H. Hu1se, H. Fow1er, & W. Honig (Eds.), Cognitive

processes in anima] behavior. Hi1]sda1e, New Jersey: Lawrence

 

Er1baum Associates, 1978.
Hearst, E., & Jenkins, H.M. Sign-tracking: The stimu]us-reinforcer

re]ation and directed action. Austin, Texas: Psychonomic Society,

 

1974.
Herbert, M.J., & Harsh, C.M. Observationa] ]earning by cats. Journa]

of Comparative Psycho]ogy, 1944, 22, 81-95.

 

Jacobson, J.M., & Sisemore, D. Observationa] ]earning of a ]ever pressing

task. Southern Journa] of Educationa] Research, 1976, 19, 59-73.

 

52

Jacoby, K.E., & Dawson, M.E. Observation and shaping: A comparison

using Long Evans rats. E§ychonomic Science, 1969, 12, 257-258.

 

Kamin, L.J. Predictabi1ity, surprise, attention and conditioning.

In B.A. Campbe11, & R.M. Church (Eds.), Punishment and aversive

 

behavior. New York: App1eton-Century-Cr0fts, 1969.
Kohn, 8., & Dennis, M. Observation and discrimination ]earning in the

rat: Specific and nonspecific effects. Journa] of Comparative

 

and Physio]ogica1 Psycho]ogy, 1972, 22, 292-296.

 

Lajoie, J., & Bindra, D. An interpretation of autoshaping and re]ated
phenomena in terms of stimu]us incentive contingencies a]one.

Canadian Journa] of Psycho]ogy, 1976, 22, 157-173.

 

Lash1ey, K.S., & McCarthy, D.A. The surviva] of the maze habit after

cerebe11ar injuries. Journa] of Comparative Psycho]ogy, 1926, 2,

 

423-433.
Lieberman, D.A., McIntosh, D.C., & Thomas, G.V. Learning when reward

is de]ayed: A marking hypothesis. Journa] of Experimenta]

 

Psycho]ogy: Anima] Behavior Processes, 1979, 2, 224-242.

 

Locurto, C.M., Duncan, H., Terrace, H.S., 8 Gibbon, J. Autoshaping in
the rat: Interposing de]ays between responses and food.

Anima] Learning and Behavior, 1980, 2, 37-44.

 

Mackintosh, N.J. Stimu]us se]ection: Learning to ignore stimu]i that
predict no change in reinforcement. In R.A. Hinde, & J.S.Hinde

(Eds.), Constraints on ]earning, New York: Academic Press, 1973.

 

Me11gren, R.L., & O]son, M.W. Associative processes contro]]ing the
persistence of operant responding: S - S* and R - S*. Bu11etin

of Psychonomic Society, 1980, 12, 279-282.

 

53
Moore, 8.R. The ro]e of directed Pav1ovian reactions in simp1e
instrumenta] ]earning in the pigeon. In R.A. Hinde, & J.S. Hinde
(Eds.), Constraints on ]earning. New York: Academic Press, 1973.

Myers, W.A. Observationa] ]earning in monkeys. Journa] of the

 

Experimenta] Ana1ysis of Behavior, 1970, 15, 225-235.

Siege], 5. Nonparametric statistics for the behaviora] sciences.
New York: McGraw-Hi11, 1956.

S1onaker, J.R. The effects of pubescense, oestration, and menopause

on the vo1untary activity of the a1bin0 rat. American Journa]

 

of Physio]ogy, 1925, 21, 362-394.

 

So1omon, R.L., & Turner, L.H. Discriminative c1assica1 conditioning
in dogs paraTyzed by curare can ]ater c0ntro1 discriminative

avoidance responses in the norma] state. Psycho]ogica1 Review,

 

1962, _6_9_, 202-219.

Suther]and, N.S., & Mackintosh, N.J. Mechanisms of anima] discrimination

 

]earning. New York: Academic Press, 1971.

Thorndike, E.L. Anima] inte11igence: Experimenta] studies. New York:

 

MacMi11an, 1911.
Tomie, A. Interference with autoshaping by prior context conditioning.

Journa] of Experimenta] Psycho]ogy: Anima] Behavior Processes,

 

1976, 2, 323-334.

Tomie, A., Murphy, A.L., Fath, S., & Jackson, R.L. Retardation of
autoshaping fo]]owing pretraining with unpredictab1e food: Effects
of changing the context between pretraining and testing.

Learning and Motivation, 1980, 11, 117-134.

 

54
Van Oyen, H.G., Van de P011, N.E., & de Bruin, J.P.C. Effects of
retention interva] and gonadectomy on sex difference in passive

avoidance behavior. Physio]ogy and Behavior, 1979, 22, 859-862.

 

Wa1ker, E.L. Reinforcement - "The one ring." In J.T. Tapp (Ed.),

Reinforcement and behavior. New York: Academic Press, 1969.

 

Warden, C.J., & Jackson, T.A. Imitative behavior in the rhesus

monkey. Journa] of Genetic Psycho]ogy, 1935, 22, 103-125.

 

Wasserman, E.A. Autoshaping: The se]ection and direction of behavior

 

py predictive stimu]i. Unpub1ished doctora1 dissertation,

 

Indiana University, 1972.

Watson, J.B. Imitation in monkeys. Psycho]ogica1 Bu11etin, 1908, 2,

 

169-178.
Wesse]]s, M.G. The effects of reinforcement upon the prepecking

behaviors of pigeons in the autoshaping experiment. Journa] of

 

the Experimenta] Ana1ysis of Behavior, ]974, 21, 125-144.

 

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

111111111)1111111111111111111111111I