VISUAL CUES USED IN ORIENTATION BY WHITEFOOTED MICE, PEROMYSCUS LEUCOPUS: A LABORATORY STUDY Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY JAMES K. JOSLIN .1 9 7 l W LIBRARY Michigan Sure University This is to certify that the thesis entitled VISUAL CUES USED IN ORIENTATION BY WHITEFOOTED MICE, PEROMYSCUS LEUCOPUS: A LABORATORY STUDY presented by JAMES K. JOSLIN has heen'accepted towards fulfillment of the requirements for Ph.D. degree in Zoolozv /\ /~ \\ “L I C\\ IN LEI Major professor Date September 13, 197_l M-795 ABSTRACT VISUAL CUES USED IN ORIENTATION BY WHITEFOOTED MICE, PEROMYSCUS LEUCOPUS: A LABORATORY STUDY BY James K. Joslin The orientation behavior of male whitefooted mice (Peromyscus leuc0pus) was assessed in the laboratory by examining their preferences for visual cues, which were analogous to ones in nature. Preferences for objects, spatial location, and object versus spatial location were examined. These preferences provided evidence for a hier— archy of factors which may influence orienting tactics of these rodents, both in the laboratory and in their woodland habitat. A water—deprived mouse was trained to exit from a tunnel in the center of a four-foot diameter (116 cm) circu— lar arena and to select one of eight levers along the arena's periphery. The correct position was indicated by two visual cues. On making the correct response, the mouse returned through the tunnel to a cage below the arena for its water reinforcement. Every second trial the cues were randomly moved to a new position. James K. Joslin After a mouse learned this task, its preferences were tested by placing the two cues 180o apart from each other. Preference was assessed in terms of the number of choices of each cue. Six mice were tested on nine different one sets, with two cues per set. The object cues were a verti- cal tube, horizontal tube, gray bowl, and luminous circle. Spatial location cues involved an object placed by the correct box, above it, 45° from it, or in front of it. A hierarchy of orientation preferences for various factors, arranged in descending order, was obtained: 1) lack of parallax (a mouse oriented to a one that was unambiguously associated with the correct box regardless of his position in the arena); 2) verticality; 3) horizontality; 4) contrast with background; 5) spatial contiguity, shape and size (all seem to have about the same rank in the hier- archy); 6) luminosity; 7) shadows. The results imply that in its woodland home range g. leucopus orients primarily to trees proximally associated with its goal (food, nest site, etc.) but when necessary can use trees 450 from the goal. The mice may use fallen logs infrequently to orient by, rarely use mounds of earth or rocks, and never use the moon. VISUAL CUES USED IN ORIENTATION BY WHITEFOOTED MICE, PEROMYSCUS LEUCOPUS: A LABORATORY STUDY BY I: A L‘ e: James K} Joslin 1"“ A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Zoology 1971 ACKNOWLEDGEMENTS I wish to thank John A. King, my advisor, for his criticisms during all stages of this project. I also wish to thank M. Balaban, R. Baker, and R. Raisler for their helpful advice in editing this manuscript. Thanks are due to my wife, Dottie, for her invaluable help throughout this study. Finally, I thank my fellow graduate students for their suggestions and constructive criticisms. This research was supported by NIH training grant 1 TOl GM—Ol751, a part of the Animal Behavior Program in the Department of Zoology at Michigan State University, and by research grant 5 R01 EYO447 from the National Eye Institute to Dr. John A. King. ii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . LIST OF FIGURES. . . . . . . . . . . . . . . . . . INTRODUCTION . . . . . . . . . . . . . . . . . . . LITERATURE REVIEW. . . . . . . . . . . . . . . . . Homing Studies. . . . . . . . . . . . . Home Definition, Basic Methods, and Dependent Variables . . . . . . . . . . . . . General Environmental Effects on Homing. Homing and the Characteristics of Indi- viduals . . . . . . . . . . . . . . Homing and Social Interactions . . . . . Homing and Initial Orientation . . . . . Homing and Randomness of Movements . . . Mechanisms for Nonrandom Homing. . . . . Cues and Homing by Piloting. . . . . . . Conclusions on Homing Studies. . . . . . Range Orientation. . . . . . . . . . . . Evidence that Peromyscus Orients within the Home Range. . . . . . . . . . . Possible Functions of Peromyscus Home Range Orientation . . . . . . . . . Sensory Modalities Used in Short Distance Orientation . . . . . . . . . . . . Peromyscus Sensory Modalities and Orien— tation. . . . . . . . . . . . . . . Home Range Orientation and Visual Cue Preferences . . . . . . . . . . . . Conclusions on Home Range Orientation. . iii Page vii 14 17 18 23 24 26 27 29 31 34 37 39 TABLE OF CONTENTS——continued PILOT STUDIES. . . . . . . . . . . ; . . . . . . First Pilot Study . . . . . . . . . . . . . Second Pilot Study. . . . . . . . . . . . . Third Pilot Study . . . . . . . . . . . . . METHODS AND MATERIALS. . . . . . . . . . . . . . subjects. . . . . . . . . . . . . . . . . . Apparatus . . . . . . . . . . . . . . . . . Early Training Stages. . . . . . . . . Large Arena Stages . . . . . . . . . . Procedures. . . . . . . . . . . . . . . . . Early Training Stages. . . . . . . . . Large Arena Stages . . . . . . . . . . Statistical Analysis. . . . . . . . . . . RESULTS. . . . . . . . . . . . . . . . . . . . . Object Preferences. . . . . . . . . . . . . Spatial Location Preferences. . . . . . . . Object Preferences versus Spatial Location Preferences. . . . . . . . . . . . . . Test Data Other than Cue Preference Data. . DISCUSSION . . . . . . . . . . . . . . . . . . . SUMMARY. . . . . . . . . . . . . . . . . . . . . BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . iv Page 41 41 42 45 58 59 6O 60 61 67 69 78 81 81 85 92 92 106 127 129 1... “gm; - TABLE 1. 10. 11. LIST OF TABLES Visual (V) versus tactile (T) cue preference in two subspecies of Peromyscus maniculatus. Peromyscus cue choices for opposition trials in the third pilot study . . . . . . . . . . Cue sets used in the experimental study. . . . Number of trials the underlined object was chosen by individual mice when objects were opposed. . . . . . . . . . . . . . . . . . . Test errors and latencies for object pref— erences: overall medians and ranges . . . . . Number of trials a particular displacement of a vertical tube (underlined) was chosen by individual mice when displacements were opposed. . . . . . . . . . . . . . . . . . . . Test errors and latencies for spatial displace— ment preferences: overall medians and ranges. Number of trials the underlined cue set was chosen by individual mice when object pref— erence was Opposed to spatial displacement preference . . . . . . . . . . . . . . . . . Test errors and latencies for object pref- erences versus spatial displacement preferen— ces during opposition trials . . . . . . . . . Number of errors (medians and ranges) in sequential control and routine trials. . . . . Latencies in seconds (medians and ranges) for the first lever pressed in sequential control and routine trials . . . . . . . . . . . . . . Page 46 54 75 84 86 89 91 95 96 100 101 LIST OF TABLES——continued TABLE 12. One tailed probability values for comparison of median errors in control versus remOval and opposition trials. . . . . . . . . . . . . . . 13. Percentages of total number of trials (N = 228) on which a particular number of errors was Obtained O O O O Q 0 O O 0 O 9 I O O O O O O 0 14. Summary of evidence for a preference hier- archy. . . . . . . . . . . . . . . . . . . . . vi Page 103 105 118 LIST OF FIGURES FIGURE Page 1. Cues used in the third pilot study. . . . . . 51 2. Cue locations in the arena for the third pilot study . . . . . . . . . . . . . . . . . 53 3. Apparatus used in early training stages . . . 63 4. Front View of large arena . . . . . . . . . . 64 5. Means and standard errors of object prefer- ences on opposition trials. . . . . . . . . . 83 6. Means and standard errors of cues displaced along various spatial coordinates on opposi- tion trials . . . . . . . . . . . . . . . . . 88 7. Means and standard errors of object versus spatial displacement preferences on opposi— tion trials . . . . . . . . . . . . . . . . . 94 8. Total number of opposition trials on which a box was chosen. . . . . . . . . . . . . . . 98 vii INTRODUCTION Rodent orientation within the home range has been ob— served during many field studies. For instance, deermice (Peromyscus) frequently ran directly to trees, logs, holes,' rocks, and other refuge sites on live trap release. Familiarity with the home range enables a mouse to locate food sources, escape from predators, and maintain nest sites and social organization (Stickel, 1968). Although deermice (and other rodents) spend most of their lives inside the home range, orientation studies have, nevertheless, concentrated on their ability to return when displaced outside the home range (Stickel, 1968). Most rodent "homing" is probably based on nonrandom movements in familiar areas outside the home range and random wandering when in unfamiliar areas. The cues used in such orienting are essentially unknown (see literature review). The relatively few laboratory studies on rodent orien— tation have concentrated on the sensory modalities used in orientation. Rats have been most extensively studied and vision seems to be the predominant modality used in mazes where the surroundings are visible (Munn, 1950). Vision also seems to be the predominant modality used by hamsters in orienting to a row of boxes (Brotzler, 1963). These conclusions are somewhat suspect, however, for much of the evidence involves surgical intervention (rats) or bright light illumination (hamsters). The sensory modalities or cues used in orienting by Peromyscus, another nocturnal rodent, have not been studied although some indirect evidence suggests that vision might be used (King, 1970; Sheppe, 1965a; Smith and Speller, 1970). Since the cues used in rodent home range orientation are unknown, this laboratory study investigated visual cues that Peromyscus may use in moving about among familiar ob— jects in its home range. Analogs of natural objects illum- inated at natural light levels provided the cues which the mice could simultaneously select in their orientation to a goal. It was hypothesized that both object and spatial location preferences exist in a hierarchical order and that one of these two perceptual factors takes precedence over the other in the orientation of mice. LITERATURE REVIEW This literature review illustrates certain concepts in small mammal orientation. With the major exception of rat maze studies, most rodent orientation studies have dealt with long—distance orientation (i.e., orientation outside the home range). These homing studies will be considered first. Then, studies dealing directly or indirectly with short—distance orientation (orientation within the home range) will be considered. Homing Studies Definition, Basic Methods, and Dependent Variables Most homing studies have involved releases of rodents at distances at least several times greater than their home range diameter. ("Home range" is that area within which the animal spends most of its time.) Homing is considered to have taken place if the rodent is subsequently observed (usually via live trap capture) to have returned to some portion of its home range (Stickel, 1968). The basic dependent variables in homing studies are the percent of subjects homing at various displacement distances (roughly, distance from the purported home range) and the time taken to home (usually measured in days) as a function of the displacement distance. Typically, the percent of subjects homing is inversely related to the displacement distance in field studies of Apodemus silvaticus, A. flavicollis, and Clethrionomys glareolus (Lehmann, 1956; Bovet, 1962); Peromyscus leuc0pus (Burt, 1940), P, maniculatus (Murie and Murie, 1931; Murie, 1963), g. gossypinus (Griffo, 1961); Microtus pennsylvanicus (Robinson and Falls, 1965); Reithrodontomys megalotis (Fisler, 1966); and Tamiasciurus hudsonicus (Layne, 1954). No relationship between distance and homing success was found for P, maniculatus according to Rawson (1966); however, he provided no data. Displacement distances in these various studies were usually somewhere within the 70-1600 yard range, with percent homing generally varying between 85% and 0%. Diverse hypotheses on the mechanisms underlying this inverse relationship between homing success and distance have been offered. The most frequent hypothesis asserts vary— ing degrees of familiarity with areas around the home nesting site. For example, Robinson and Falls (1965) stated that at short distances voles (Microtus) have complete familiarity with the release area, at intermediate distances they have less familiarity (fewer voles have knowledge of the area and also recognize fewer landmarks), and at long distances they have no familiarity with the release area. Some of the evidence they cited was the poorer homing success for enclosurewreared voles as compared to controls when both were field—released at the same distance. Also, their literature survey revealed that the maximal displacement distance from which a small terrestrial mammal will home is directly related to the home range size. Differential familiarity with an area and observed per- cent homing success-have been explained on the basis of 1) the home range itself being a series of concentric prob- ability-of occurrence zones (Robinson and Falls, 1965), 2) the home range itself being under—estimated by various live-trapping methods (Chitty, 1937; Robinson and Falls, 1965), 3) shifting of home range by an individual (Griffo, 1961), 4) exploratory trips outside the home range (Stickel and Warbach, 1960; Griffo, 1961), and 5) distribution of juvenile dispersal distances outside the parents' home range (Griffo, 1961). The inverse relationship has also been explained on the basis of probability of success at varying distances for random search (Rawson, 1966; Robinson and Falls, 1965). Several random search models were reviewed by Robinson and Falls (1965). They did not find that such models helped in the interpretation of their data. Other possible interpretations of the inverse relation— ship include the following: 1) Increased opportunity for encountering favorable (and presumably unoccupied) habitat may occur as a function of displacement distance. This may lead to the founding of a new home site (Hamilton, 1937; Bovet, 1962). 2) Increased probability of individual mor— tality through predation may occur as displacement distance increases (Bovet, 1962). However, Robinson and Falls (1965) noted that for their voles, practically all individuals returned on second releases, regardless of displacement distance. 3) An assumed constant angular error of orienta— tion may be present, which for greater distances would result in a mouse's having less chance of encountering its home range (Bovet, 1962). Homing times for a given distance have been typically highly variable for homing experiments. For example, Fisler (1962) recorded a range of 1—9 days for 200 foot displace— ments in voles; Murie (1963), 1-8 nights for 800 yard dis- placement in deermice; and Fisler (1966), 1—47 days for 320 foot displacement in harvest mice. Fisler (1962) has sug— gested that in Microtus californicus these individual dif- ferences could be attributed to such variables as physio- logical conditions, social and psychological motives, wandering tendencies, and dispersal tendencies. Some of this variability could be accounted for on the basis of Robinson and Falls' (1965) postulated individual differences in familiarity with the surrounding area, or perhaps by sub— jects not entering live traps for some time after returning to the home range. The latter possibility was suggested by radio—telemetric tracking of an individual Peromyscus maniculatus (Rawson and Hartline, 1964). Homing time was not related to displacement distance (Schleidt, 1951; Murie, 1963; Robinson and Falls, 1965; Fisler, 1962; and Fisler, 1966). This can be attributed partly to the individual variability in homing time discussed above, and perhaps partly to the rather crude way time has been measured (in nightly or daily intervals). The crude- ness of this measurement was illuminated by Rawson and Hartline (1964). Using radio—telemetry they obtained a homing speed of 300 m./hr. for Peromyscus maniculatus as compared with the usual live trap studies' speed of about 30 m./hr. One of their mice returned home but did not enter a live trap the first night. However, crudeness of the measurement alone probably can not account for the lack of a relationship between homing time and distance displaced. A considerable degree of individual variability in return time also existed in a live trap study using two hour time intervals (Gentry, 1964). General Environmental Effects on Homing Variability in homing performance can be related not only to individual variability but also to general environ— mental effects. A directional movement bias in homing has been noted by Bovet (1962). Apodemus flavicollis and A. silvaticus were caught in traps located 250—1000 meters from the release point significantly more often in the sou— therly direction, to an intermediate extent along an east— west axis, and least in the northerly direction. This effect was not explicable in terms of the individual's home site directions or in terms of predation. A southward movement bias may also be present for Peromyscus leucopus, as sug— gested by some of Burt's 1940 data. Bovet concluded that these directional differences imply that the mice were orienting "according to [unspecified] factors they perceived at the release site." However, in both Bovet's and Burt's study areas there was some habitat heterogeneity, perhaps enough to bias movement patterns. Directional differences in homing have also been interpreted in terms of differential familiarity with landmarks (Davis, 1966). This, in turn, could be based on habitat hetero— geneity. Indeed, Bovet (1965a) himself has some evidence for the effects of habitat on homing success. Apodemus silvaticus homed with greater success from a zone relatively poor in vegetation and topographical features as compared to a zone rich in vegetation and topographical features. Habitat effects on homing success have also been noted for Peromyscus by Murie (1963) and Griffo (1961); for Microtus by Robinson and Falls (1965) and Fisler (1962); for Reithrodontomys by Fisler (1966); and for Neotoma by Lay and Baker (1938). Habitat effects on homing success have been interpreted on the basis of poor habitat in one direction from the release site being a barrier to movement in that direction (Murie, 1963), poor habitat at the release site not offer— ing enough concealment which in turn facilitates phycho- logical tendencies to return home (Fisler, 1966), and poor habitat not having familiar cues to orient by (Griffo, 1961; Robinson and Falls, 1965). Seasonal effects have been implicated for Apodemus silvaticus, with this species homing better in winter than in summer. Winter subjects may have accumulated greater familiarity with surrounding areas and may also have been less likely to establish a new home site rather than home to the old one (Lehmann, 1956). Other factors, such as fluctu- ating barometric pressures, may also be involved, for this species also exhibits better initial orientation (homing direction) in a laboratory maze in winter than in summer (Bovet, 1960). This winter effect may be species specific, for Mus musculus was disoriented in its initial maze orien— tation in winter (Lindenlaub, 1955). Homing and the Characteristics of Individuals Although general environmental influences may be present, characteristics of individuals have received greater empha— sis in homing studies. Improvement of an individual's homing performance (practice effects) on subsequent releases at the same distance has been frequently noted. Such a practice effect was present even when subsequent releases were made in 10 different directions. Peromyscus gossypinus exhibited more direct initial orientation movements toward the home site on its second release in an unnatural habitat, a golf course (Griffo, 1961). A decrease in homing time for the second release has been demonstrated for Apodemus silvaticus (Bovet, 1965a), Peromyscus maniculatus (Murie, 1963), Peromyscus gpssypinus (Griffo, 1961), Micurotus pennsylvanicus (Robinson and Falls, 1965), and Microtus californicus (Fisler, 1962). An increase in the percent of subjects homing on subsequent releases has been demonstrated for Peromyscus maniculatus (Rawson, 1966; Murie, 1963), Peromyscus gossypinus (Griffo, 1961) and Microtus californicus (Robinson and Falls, 1965). However, Bovet (1968) found no practice effect for Peromyscus maniculatus; but his second release sample size was very small. A weak practice effect, at best, was present for Reithrodontomys megalotis (Fisler, 1966). However, about half of Fisler's subjects were displaced twice as far on the second release. This practice effect has been interpreted as due to an animal's acquiring some familiarity with a large area during its first release movements. On second release the subject would be at least partially familiar with that area (Griffo, 1961; Murie, 1963; Robinson and Falls, 1965). A good memory would be necessary, of course. Two studies demonstrated such a memory. Peromyscus gossypinus was retained in the labora— tory for about one month after some homing experience, with 11 no subsequent decrease in homing ability when released at prior release sites (Griffo, 1961). Peromyscus maniculatus individuals, although exhibiting no significant initial orientation homeward, did show a highly significant tendency to follow the same initial route on second displacement (Murie, 1963). Males frequently exhibited a higher percent homing suc— cess than did females for various displacement distances. This has been shown for Peromyscus leucopus (Stickel, 1949), Peromyscus maniculatus (Murie, 1963), Peromyscus gossypinus released from a natural habitat (Griffo, 1961), Microtus californicus (Fisler, 1962), and perhaps for Microtus _pennsv1vanicus (male percent homing success was greater for eight out of twelve displacement distances, Rdbinson and Falls, 1965). No sex effect was present for g, gossypinus released in an unnatural habitat (Griffo, 1961), and sex effects were minimal at best for P, polionotus (Gentry, 1964). Pregnancy itself did not necessarily interfere with female homing (P, maniculatus, Murie, 1963; and Reithrodontomys megalotis, Fisler, 1966). Greater male success in homing has been explained on the basis of males having familiarity with a larger area than do females. Reasons given were that males had larger home ranges (Stockel, 1949; Griffo, 1961), males made longer exploratory trips outside their home ranges (Murie, 1963; Griffo, 1961), and juvenile males dispersed to longer distances (Griffo, 12 1961). In an unnatural habitat release area (a golf course) which was presumably a totally unfamiliar area, Griffo (1961) obtained no sex differences, in contrast to releases in a natural habitat. This difference adds credence to the dif— ferential familiarity hypothesis for sex differences in homing. In several studies, adult homing success has been greater than juvenile success (Peromyscus maniculatus, Murie, 1963; Microtus pennsylvanicus, Robinson and Falls, 1965; and Reithrodontomys megalotis, Fisler, 1966). This has been attributed to juveniles having less familiarity with surround- ing areas (Griffo, 1961) and to juveniles lacking the motiva- tion to return, perhaps because they have not yet established a home range of their own (Griffo, 1961; and Murie, 1963). Direct evidence for juveniles being less motivated to return was suggested by Murie (1963). Eighty-five percent of adult and subadult Peromyscus maniculatus, as compared with sixty— five percent of the juveniles, were caught the first night in live traps arranged circumferentially about one hundred feet from the release point. An alternative explanation, however, for the Murie (1963) results would be that juveniles are less likely to enter traps outside the home site area than are adults after displacement in a homing experiment. Although no such specific age—related effect has apparently been demonstrated, it has been shown that some small mammals do not readily 13 enter live traps placed outside their home range during hom- ing displacement experiments (Peromyscus maniculatus, Morris, 1967; Peromyscus gossypinus, Griffo, 1961; and Microtus pennsylvanicus, Robinson and Falls, 1965). Even the Murie (1963) results might illustrate this phenomenon since all displaced mice before their displacement were caught inside their home ranges (presumably) while eighty—five to sixty— four percent were caught outside the home range. Griffo (1961) has interpreted this decrease in trap— pability outside the home range as being related to psycho- logical and perhaps physiological stresses, which result in a search for the familiar area and an avoidance of live traps. Within the home range, the animal's "psychic" needs are met, it moves with "assurance", is not under stress, and enters live traps. There is no direct evidence for such postulated stresses on subjects displaced outside their home ranges in homing experiments. Furthermore, either some individuals were not particu- larly stressed by displacement or they recovered from such a stress rather quickly. This was indicated by reports of displaced adults establishing new home ranges, often near the release site, rather than homing (Peromyscus: Murie, 1963; Griffo, 1961; Rawson and Hartline, 1964; Gentry, 1964; Stickel, 1949; Microtus: Robinson and Falls, 1965; Hamilton, 1937; and perhaps Reithrodontomys of unspecified age(s): Fisler, 1966). subsequent homing releases of such individuals have resulted in their homing to the "new" home rather than the 14 "old" one (Murie, 1963; Griffo, 1961; Fisler, 1966). However, lack of homing and establishment of a "new" home site on displacement may not be the only interpretation for the observed live trap phenomena. Fisler (1966) has sug- gested that for Reithrodontomys the home ranges shift fre— quently enough that a displaced mouse is really homing, but to a previous home range. Such a postulated displacement stress could well oper— ate only during part of a year. In Apodemus silvaticus, percent homing success decreased as the mice became less sedentary (Bovet, 1965a). A similar effect has been noted for Peromyscus maniculatus (Bovet, 1970), perhaps related to interspecific interactions. 3, maniculatus became less sedentary about the same time that the first active ground squirrels were seen around the granaries Peromyscus was liv- ing in. Homing and Social Interactions Intraspecific interactions have been implicated more often than have interspecific interactions. Social inter- actions with conspecifics at the release site can facilitate homing, perhaps by increasing the stress a small mammal may be under (Fisler, 1962; Griffo, 1961). Displaced Microtus californicus remained at a release station which lacked resident voles, in contrast to another release station at the same distance which had resident voles (Fisler, 1962). A similar effect has been noted for Adppdemus silvaticus 15 (Bovet, 1965a and 1968). However, no such social effect has been shown by Reithrodontomys megalotis (Fisler, 1966). A similar homing facilitatory effect seemed to be present for points intermediate between release site and home site (Bovet 1965a; Stickel, 1949). Such-social effects on homing may be due to a tendency for individuals to mutually avoid each other (Murie, 1963; Terman, 1962). .However, such interactions are not abso— lutely necessary for homing to occur. Stickel (1949) found that most Peromyscus leucopus homed even though they were moving through areas with all conspecifics removed. At the home site itself, the presence of a conspecific can inhibit homing. Peromyscus maniculatus adults homed significantly less to their nest boxes when a young alien mouse was tethered by a nest box. Mutual avoidance has been suggested as the basis for this decreased homing (Terman, 1962). Homing and Initial Orientation Besides social and other variables which can influence homing success, homing movements themselves have been ana— o lyzed. Initial orientation in the homeward direction, i 90 from the home site, has been exhibited by Peromyscus manicu— latus released on winter snow from 100—500 meters displace- ment distances (Bovet, 1971). Average angular errors from “the homeward direction were no greater than 370 for 20 m. to 5“? m. distances from the release site (Bovet, 1968). 16 Peromyscus maniculatus also exhibited significant homeward orientation, : 90°, in another field release study (Rawson, 1966), but no data was provided for evaluation. Most field release studies, however, indicate a lack of initial orien- tation in the homeward direction for various rodents (Peromyscus: Murie, 1963; Griffo, 1961; Gentry, 1964; Microtus: Rdbinson and Falls, 1965; and Apodemus: Lehmann, 1956). Bovet (1968) has suggested that his positive results were due to heightened motivation (his mice were released on an open snow-covered plain, with low temperatures) and to lack of "diverting" cues (such as cover and food) near the release site. This postulated cover-seeking tendency is supported by direct observations on released Peromyscus polionotus (Gentry, 1964). Initial orientation in the homeward direction has also been demonstrated in laboratory studies of Eur0pean rodents. Releases were made in the center of a labyrinth maze with 18-24 exits along the periphery. Displacement distances from the home site varied between 70 m. and 3.7 km. The initial homing orientation seemed rather weak, though, for statistical significance was obtained in only 6 of 54 experi— mental series (Bovet, 1960), or by pooling the data of several experimental series (Lindenlaub, 1955). No initial homing orientation was present for two North American rodent Species released from the center of a circular platform placed it) a grassland (Fisler, 1967). 17 According to Bovet (1971), Fisler's negative results could be due to habituation (Fisler's subjects received ten trials in 24 hours; Bovet's and Lindenlaub's, one), seasonal effects, and the presence of distracting habitat and celes- tial cues which could elicit cover—seeking tendencies instead of initial orienting. Homing and Randomness of Movements Conclusions on randomness of movements in homing studies usually have been based on intermediate and final point orientation rather than on initial orientation. Most workers have concluded that their subjects' homing performance was either totally nonrandom (Peromyscus: Burt, 1940; Bovet, 1968; Rawson, 1966; Apodemus: Bovet, 1962; Reithrodontomys: Fisler, 1966) or was at least partially nonrandom (Peromyscus: Griffo, 1961; Morris, 1967; Murie and Murie, 1931; Clethrion— megs Schleidt, 1951; Apodemus: Lehmann, 1956; Microtus: Robinson and Falls, 1965; Fisler, 1962; Tamiasciurus: Layne, 1954; Tamias: Layne, 1957). The minority View of homing being accountable simply on the basis of randomness has been offered by Murie (1963). Conclusions on the partial or complete nonrandomness of homing have generally been based primarily on homing speeds and percent homing success, and secondarily on practice effects and direct observations of movements. Murie's conclu- sion of complete randomness apparently was based mostly on lack of initial homeward orientation by his mice. This could 18 be explicable on the basis of cover—seeking tendencies being predominant (Bovet, 1968). The evidence from these studies clearly implies that homing movements are at least partially nonrandom. Direct snow tracking evidence (Bovet, 1968 and 1971) and comparison of performances in natural and unnatural habitats (Griffo, 1961) are particularly impressive evidence in favor of a nonrandom component. Practice effects and positive evidence cited for initial orientation also strongly imply a non- random component. Homing from various displacement dis— tances based completely on random search activities would be most likely for a species having very small home ranges, extremely slight tendencies to wander outside the home range, very small dispersal distances, and extremely poor distance receptors (vision and hearing). Not many rodent species could meet all these criteria! Mechanisms for Nonrandom Homing What mechanisms might be involved in nonrandom homing movements? Random search activities outside a familiar area combined with directed movements using cues present in a familiar area (piloting) has been frequently cited as a mechanism underlying rodent orientation. This topic has been discussed previously. Such a mechanism seems highly plausible. Systematic search activities (for example, in the form of a spiral) combined with piloting in familiar areas have 19 been suggested for some birds by Griffin and Hock (1949). The evidence for such a mechanism in birds is scant (Schmidt- Koenig, 1965) and there is apparently no evidence for such a mechanism in rodents. A priori, such a mechanism seems unlikely since in itself it would probably require a high degree of orientation, perhaps even navigation. A more goal— directed orientation mechanism would probably be selected for during the course of evolution, given the same type of selection.pressure. Navigation of some sort is another possible mechanism for rodents displaced during homing studies. Navigation is defined as the ability to sense one's exact geOgraphical position on displacement, know the "map" co-ordinates of the home site, and then move along a rather direct course between these two known points. Navigation could be based on pro- prioceptor input (inertial navigation), or information ac— quired from celestial bodies' locations, or from geophysical gradients (e.g., magnetic gradients). Navigation has been suggested for the home—directed winter releases of Peromyscus maniculatus (Bovet, 1968); and "homing instinct" (presumably based on navigation) has been suggested by Murie and Murie (1931) and Burt (1940). The laboratory experiments of Lindenlaub (1955 and 1960) and Bovet (1960) on initial orientation could suggest a navigational ability in some European rodents. However, as previously noted, the statistical evidence was rather weak. 20 Furthermore, Lindenlaub's results may be explained by pos— sible orientation to some olfactory, auditory, or other cue associated with the nearby sea, rather than to navigational ability (Bovet, 1960). Although Bovet's experiments were not near the sea, perhaps a similar type of explanation applies to his results. Furthermore, Bovet (1962) maintained that there is a constant angular error in navigation, but his subjects from one area near the laboratory exhibited poorer performance than those from an area further away (Bovet, 1960). It is questionable how valuable a navigational mech— anism would be that is affected adversely by so many variables as Bovet‘s Table 2 suggests. Odor traces left in his appa— ratus may also have influenced results. Bovet (1965b) was able to train, to some extent, Apodemus silvaticus to choose various directions in a maze. This rather weakly demonstrated ability might not be based on navigational mechanisms, however. Placement of elements in the experimental room was heterogeneous enough to have perhaps allowed for echolocation. The winter homing study of Peromyscus maniculatus (Bovet, 1968) does not necessarily imply navigation, although this mechanism seemed most likely to Bovet. Bovet's data does not suggest to me a mechanism as precise as navigation for the following reasons: 1) there was only a 70% homing success rate, and 2) actual movements of the mice involved large distance zig—zagging, some extensive retracing of routes, and 21 a traversing of a; legs; twice the straight-homeward line distance (only part of the total homeward movement could be followed). Instead, it seems more likely that Bovet's mice were orienting to partially familiar cues in the area of their home granary. Although these granary mice were sedentary during the winter, they became less sedentary toward the end of March (Bovet, 1970). Familiarity with the surround- ing areas could have been gained during the previous spring and summer. Such familiarity could be retained by memory processes (such as those demonstrated by Griffo, 1961). Indeed, Lehmann,(l956), has suggested that his mice homed better in winter because of greater prior acquired familiarity with surrounding areas. Bovet's displacement distances certainly did not pre- clude such familiarity. His mice were successively released at 100—200-300—400—500 meter distances. Various evidence from Stickel's (1968) review article, particularly as it ap- plies to grassland subspecies of Peromyscus maniculatus, supports the possibility of such familiarity, namely: 1) Home ranges had radii (or observed length/2) of 37—117 meters. Thus, a mouse displaced at 100 meters could be familiar with such an area already and when displaced at 200 meters orient to familiar cues present in the 100 meter area, concurrently gaining some experience with 200 meter zone cues, etc. 2) Juvenile dispersal distances were frequently 22 in the 180—340 meter range. 3) Adult movements of at least 340 meters outside the home range are known to occur. Certainly there were sufficient cues present for Bovet's mice to be familiar with and orient to, such as: the gran— aries themselves, wheat field stubble in part of the area, straw piles, fences, small hill crests, various electric lights in farms, and glow from the lights of a nearby city. Bovet discounted such luminous cues but provided no data to support his claims. Kinesthetic cues during transport to the release area are another possibility. Direct perception of home site stimuli might be another homing mechanism (Bovet, 1968). Evidence positive or nega— tive is not plentiful for such a postulated mechanism, but it could account for the homing of three Peromyscus gossypinus to home site areas near noisy pigpens (Griffo, 1961). Another homing mechanism would involve the following of some gradient (such as social odors) to the home site (Bovet, 1968). Evidence for such a mechanism is apparently lacking, and Bovet (1968) discounted an olfactory gradient mechanism operating in his study because of "the complete absence of any Peromyscus sign along a radius of at least 30 m. around any release point used." Familiarity with terrain cues and piloting with refer— ence to these cues has been frequently suggested as a mechanism, alone or in conjunction with a random search phase. This highly plausible and simple mechanism has already been discussed in some depth. 23 Cues and Hominquy Piloting Considerably less is known about what cues a rodent might use in homing than there is about what perceptual mechanisms might be involved. The evidence, scant and im- perfect as it is, does suggest that at least in piloting the visual modality may play a major role. Visual cues provided by the terrain may be used in piloting. Practice effects (reviewed earlier) could imply use of such cues, as could the lower percent homing success of enclosure—reared voles on release outside the enclosure (Robinson and Falls, 1965). Also, one Peromyscus gossypinus may have oriented to the heavy shadows of a wooded section near its golf course release site (Griffo, 1961). Horizon cues were apparently not used by Reithrodontomys megalotis and Microtus californicus. They did not exhibit initial homing orientation from a circular platform in the presence of such cues (Fisler, 1967). Evidence for visual use of celestial cues is generally negative but rather weak. Clear moonlit nights inhibited Peromyscus gossypinus movements on golf courses (Griffo, 1961). Peromyscus can apparently home successfully under overcast skies (Rawson, 1956, cited by Griffo, 1961), and neither Reithrodontomys megalotis nor Microtus californicus showed initial orientation on release in the field from a circular platform. However, sun compass use has been demon— strated for Apodemus agrarius (Lfiters and Birukow, 1963), 24 for this species was trained at a certain time to a given compass direction with an artificial sun. When tested at various other times of day, it exhibited the appropriate shift in compass direction relative to the time of day. There is apparently no evidence for olfactory cues in piloting, but there is some evidence for orientation by Peromyscus gossypinus to auditory cues, the clanging of feed troughs (Griffo, 1961). Kinesthetic cues may not account for Microtus homing. Voles taken on indirect routes prior to release homed as well as controls (Robinson and Falls, 1965); however, no data was given. Conclusions on Homing Studies In spite of numerous homing studies there is still mostly only indirect evidence for the perceptual mechanisms and scant evidence for the cues that might be involved in homing. Nonrandom movements play a major role in homing, but there is yet no convincing evidence that this nonrandomness is based on navigation. Familiarity with a large area and use of terrain landmarks to orient by (piloting) probably account for most homing performances. However, seldom has any experimental control been exerted over the familiariza— tion process. Instead, the familiarization hypothesis has been supported on the basis of sex and age differences in homing, practice effects, and other indirect types of evidence. 25 Progress in homing research could occur by the design- ing of experiments specifically to test hypotheses about mechanisms and/or cues using the strong inference method, rather than the post hoc discussion presently so prevalent. A dependent variable more accurate than the usual 12 or 24 hour time unit for measuring homing would also be desirable. Some progress has been made by direct observations of part of the homing.movements (Bovet, 1968; Griffo, 1961). Radio— telemetric techniques may be worth further exploration, in spite of their cost and technical problems. I propose the following experimental paradigm fOr the strong inference test of homing mechanisms. Various experi- mental groups of island mice would be given varying degrees of exposure to mainland areas. (By using island mice, it is unlikely that any mice had prior familiarization with the mainland terrain.) Group A would consist of caged individuals moved every several days to points successively further from the island, with the island visible from the closest point. Group B individuals would be given several days' exposure each, only at the nearer points. Group C individuals would be given no mainland experience prior to release. Live trap lines would be positioned in all four compass directions at (say) 250 meters from the release site. These live traps would have a timing device to indicate when the trap was entered, or radio-telemetric receivers would be placed along the trap line. 26 If homing is based entirely on random search activities, all three groups would be captured with equal frequencies and with similar homing speeds at each of the four trap lines. If navigation is used, all three groups would be captured with equal frequencies and with similar homing speeds, but most individuals would be caught on the trap line facing the island. If area familiarization is involved, Group A should exhibit the best homeward direction performances, Group B should be intermediate, and Group C should be poorest. If the latter result is obtained, then the relative roles of terrain and horizon—celestial cues could be ascertained by providing to three groups successively distant exposures on the mainland before release. One group could be housed in cages with an opaque top and upper sides, allowing a View only of terrain cues. Another group could be housed in cages with an opaque lower portion, allowing a View only of horizon—~ce1estial cues. The third group could have an un— limited view. Conclusions would be based on frequency of capture on the four trap lines and on homing speeds, using logic similar to that described above. Home Range Orientation Except for rats, very few studies deal directly with the sensory modalities or cues used in short distance (home range) orientation. Yet a rodent spends most of its time within the home range. Thus, in terms of gaining better understanding 27 of a species' distribution, home range orientation cue studies should yield as much, if not greater, insight than have homing studies. Perceptual selectivity may indeed be one of the ultimate bases for a species' distribution (King, 1970). Of course, homing and home range orientation might both involve the same orienting mechanisms, modalities, and cues, particularly for rather short displacement distances. At longer distances perhaps the same mechanisms and modalities are used, but different cues. Throughout this thesis, perceptual cues will be con— sidered to consist of objects, spatial locations of objects, events, representations of objects, and coded stimuli. This is the "ecological" classification suggested by Gibson (1969). Object perception and preference will receive primary empha— sis, spatial locations of objects will receive less emphasis (less information is available), and the other items will not be considered since they are quite irrelevant to home range orientation. Evidence that Peromyscus Orients within the Home Range Orientation within the home range has been assumed in this review. Using Peromyscus as an example, what evidence is there for home range orientation? Live trap releases within the home range imply that Peromyscus does indeed exhibit home range orientation. On release from live traps, 28 Peromyscus moved directly to a nearby refuge site, usually a hole in the ground, a tree base, or a log (Blair, 1940; Burt, 1940; McCabe and Blanchard, 19507 Griffo, 1961; Smith and Speller, 1970). Individuals caught several times in a live trap took the same route to a refuge on several occa- sions (Blair, 1940; Burt, 1940; McCabe and Blanchard, 1950). In some instances, however, deermice would wander a short distance, sometimes stop, and Eh§g_move directly to a shelter (Blair, 1940; McCabe and Blanchard, 1950), suggesting initial disorientation on daytime release. Deermice released outside their home ranges wandered for considerable distances (Burt, 1940; Griffo, 1961). This contrast in orienting behavior with reference to the home range was exhibited graphically in simultaneous live trap releases of two individual P, gossypinus from one site. The mouse within its home range moved quickly and directly to shelter; the other mouse wandered slowly around the area (Griffo, 1961). Knowledge of objects within the home range and home range orientation by Peromyscus have also been demonstrated by other methods. P, polionotus soon re—opened food or nest holes covered by drifting sand (Blair, 1951), and g, manicu— latgg exhibited direct movements to a food cache hole with subsequent re—opening of the hole (McCabe and Blanchard, 1950). g. maniculatus on displacement within a large enclosure returned to one of its own several nest boxes, bypassing all 29 others (Terman, 1962). P, leuc0pus escaped from a human observer by direct movements to an enclosure pen treadle (Orr, 1959). This species quickly explored unfamiliar ob- jects (juice-can shelters) placed in.its home range, and also acquired a habit of visiting one particular live trap (Sheppe, 1966). Possible Functions of Peromyscus Home Range Orientation Peromyscus orientation within the home range is further substantiated by evidence that such movements are function- ally significant. Efficient location of food sources is one of the functions of home range orientation. 2. maniculatus on live trap release moved directly toward a food cache and subsequently dug for the food (McCabe and Blanchard, 1950). Frequency of P, leucoEus travels to various parts of a home range changed with the location of food sources, and home range shifts were induced by artificial provisioning of food (Sheppe, 1966). P. polionotus had quite direct heavily— frequented trails leading to areas where sea oats were blown down (Blair, 1951). Extent of orienting movements may be a function of food availability, with smaller home ranges present when food supply is abundant (Stickel, 1968). Efficient escape from predators is another function of home range orientation. In a large test room, a significant— ly higher percentage of resident P. leuc0pus survived in the presence of an owl predator as compared to the transient 30 survival percentage (Metzgar, 1967). On live trap release, P. maniculatus frequently ran down holes at tree bases (Smith and Speller, 1970). Orientation to refuge holes has also been suggested for other Peromyscus species (Stickel, 11968; McCabe and Blanchard, 1950). Maintenance of social organization and distribution of individuals is yet another function of home range orienta- tion. P, maniculatus released some distance from its home nesting sites returned to them, avoiding use of empty nest sites of its neighbors (Terman, 1962). Even when some P. leucopus were removed from an island, other individuals often did not quickly expand their home ranges (Sheppe, 1966). Attachment to the home range may help maintain an individual's status in the social structure and confer a selective advan- tage over aliens who enter this area (Sheppe, 1966). This selective advantage has been directly observed in P, 991197 ppgps. Several instances of residents evicting live—trap— released transients from their refuge holes were noted by Blair (1951). It seems likely that the functions of home range orien— tation discussed above may be important for the species' population through their enhancing effects on birth rates and their ameliorating effects on death rates. Thus, selec- tion pressure has probably placed a premium on the use of certain cues and sensory modalities for home range orienta- tion. 31 Sensory Modalities Used in Short Distance Orientation Literature relating to home range orientation and the role of various sensory modalities will be used selectively to illustrate points that relate to my research, namely: 1) Vision is used by nocturnal rodents in orienting. 2) Light levels used in a study of visual cue orientation should be within the limits encountered in the natural habi- tat. 3) Various other modalities have been implicated for nocturnal rodents. Thus, a study concerned with visual cues should have controls to ensure that other modalities were not being used in orienting to the cues. 4) Spatial position habits should be taken into account in a visual cue orienta— tion study. Vision seemed to be relied on more than olfaction in a study of hamster short distance orientation (Brotzler, 1963). The experimental apparatus consisted of an entrance at one end and a row of nine boxes at the other end, each with a swinging door. The reward was the opportunity to run on a large table behind the correct door, indicated by the follow- ing cues singly or in combination: visual cue, olfactory cue, or spatial location cue (numerical position in the row of doors). Sometimes the various cues were opposed by placing them by different doors. Unfortunately, most of Brotzler's experiments were con— ducted under bright light (75 watts). This apparently biased the results, for in one of his experiments the 32 percent correct choices (with only the visual cue present) decreased as the light level decreased. Thus the relative roles of visual and olfactory cues for a hamster's nocturnal orientation remain unclear. Short distance orientation has been most extensively studied in the laboratory rat. Vision seems to play a greater role in elevated than in enclosed mazes (Munn, 1950). Only elevated maze studies will be considered here since elevated mazes with their open visual field are presumably more comparable to a rat's home range environment. Munn's review will be relied on extensively since only a few maze studies dealing with sensory modalities as such have appeared since-1950 and their results do not substantially change the conclusions reached by him. Elevated maze studies have used such techniques as interchanging the maze units, surgically removing various sense organs, running rats in complete darkness, providing distinct visual cues for some groups but not for others, shifting the maze's position in the room, and providing a combination of distinct cues of various modalities which were subsequently presented singly. For elevated mazes, Munn (1950) concluded that vision seems to be the dominant sense in rat orientation. Olfac- tion is used when rats are deprived of visual input. Olfac— tion and audition play some role in maze orientation by seeing rats, but kinesthesis and touch play a negligible role. 33 Hewever, Munn's conclusions should be viewed with some skepticism. Many of the maze studies involved surgical interventions. At best, these methods can tell us only what an animal is then capable of doing, not what a normal animal would do. Surgical intervention could do more than just remove a sense organ. It could affect various other processes of the central nervous system. Furthermore, the visual sense seems to have been studied considerably more than some of the other senses reviewed by Munn. Indeed, Riley and Rosenzweig (1957) maintained that maze performances previously attributed to vision were due to echolocation. Their blinded rats were able to discriminate between an alley blocked by a barrier and an open alley. Rats with their hearing impaired performed at about chance level. However, they had no direct evidence for echolocation sound production. The same criticisms could apply to their experiment as apply to Munn's conclusions. Direct evidence for echolocation is present for shrew and tenrec orientation, involving a trained jumping response (Gould, Negus, and Novak, 1964; Gould, 1965). To what extent the modalities used in orienting by laboratory rats are similar to those used by wild rats is relatively unknown. Calhoun (1962), on the basis of natural— istic enclosure studies of wild rats, concluded that the relative use of sensory modalities for orientation is, in <1rder of decreasing use: tactile, visual, kinesthetic, and 34 olfactory. This order is quite different from the order based on Munn's review of laboratory rat studies. Spatial—position responding has been implicated for hamsters (Brotzler, 1963), for: l) The "correct" door was chosen.in complete darkness even when the odor cue was placed by another door. 2) The "correct" door was chosen more frequently than the door with the visual cue when there were four doors intervening. 3) Hamsters visited a previous— ly "correct" door before visiting a new "correct" door. A spatial memory may also be of particular value in white— tooth shrew orientation, as suggested by experimental dis— placement of young from the nest to a particular spatial locus, maternal retrieving of the young, and then subsequent removal of the young by the experimenter (Grunwald, 1969). Peromyscus Sensory Modalities and Orientation Although there is no study of Peromyscus comparable to the studies just reviewed, other types of studies may have implications for sensory modalities used in Peromyscus home range orientation. Large eyes, a characteristic feature of Peromyscus, could imply enhanced visual capabilities for this nocturnal genus (King, 1968). A.potentia1 role for vision in home range orientation is suggested by various laboratory studies: 1) Peromyscus has better visual acuity than any other rodent studied so far (18.2' of arc at 40 cm), perhaps partly due to the experimental method used (Vestal, 35 1970). 2) Very low light intensities, almost as low as one-millionth candle power per square centimeter, can be perceived (Moody, 1929). 3) Peromyscus can jump to a small object and land correctly even if the object is displaced (Horner, 1954). 4) Peromyscus can discriminate between small circle and cross forms of light (King, 1968). 5) Peromyscus usually oriented toward a light source while running in an activity wheel (Kavanau, 1968). Orientation to visual cues is also suggested by several field studies: 1) Peromyscus leucopus released 175 feet offshore from an island swam toward the island, perhaps orienting to tree tops (Sheppe, 1965a). 2) Peromyscus leucopus shelter track records increased when empty juice cans, presumably visual cues, were placed near the shelters (Sheepe, 1965b). 3) P, maniculatus usually ran directly to a nearby tree when released from a live trap (Smith and Speller, 1970). Tactile cues have also been implicated for Peromyscus orientation. For arboreal orienting, Peromyscus brought their vibrissae in contact with a non-contiguous branch be— fore attempting to cross the gap, and also gently waved the tail, making light contact with the branch traveled along (Horner, 1954). In total darkness Peromyscus still main- tained an oriented direction in running wheels (Kavanau, 1968), and still chose the artificial tree or artificial grass habitat appropriate for the subspecies (Harris, 1952). 36 Strong position habits, based perhaps on both tactile and kinesthetic cues, were present during a brightness dis— crimination task (Moody, 1929). Orienting by echolocating is conceivable, for Peromyscus can hear well into the ultrasonic range, up to 100 kHz (Dice and Barto, 1952). Communicative functions (Eisenberg, 1962) and predator detection (Dice, Barto, and Clark, 1963) may be based at least partially on audition. Short distance orientation (of 6 meters or less) toward buried seeds on the basis of olfaction rather than vision was exhibited by P, maniculatus (Howard, Marsh, and Cole, 1968). The percentage of buried seeds discovered by the mice did not differ significantly between dim light and total darkness. P, leucopus discriminated fairly well between oat— baited and empty track shelters (Sheppe, 1965b), presumably on the basis of odor. Olfaction has also been implicated by various studies on species discrimination and the inhibition and facilitation of Peromyscus reproduction (King, 1968). In summary, there is some evidence to suggest that any one or more of the sensory modalities could well be used in Peromyscus home range orientation. For reasons discussed later, my research concentrated on the visual modality. Controls and other experimental design procedures were used to ascertain whether the cues provided were really being used as visual cues and to ensure that spatial position habits could not bias results. 37 Home Range Orientation and Visual Cue Preferences What visual cues would Peromyscus probably use in home range orientation? Spontaneous visual cue choice studies of various rodents could implicate certain cues. However, there have been few spontaneous visual cue choice studies on nonhumans even though such information would be highly desirable to have before conducting a discrimination learn- ing study (Sutherland, 1961). Annual Review of Psychology articles since 1961 also suggest a paucity of spontaneous cue choice studies. Munn's (1950) extensive review of psychological research on the rat mentioned only spontaneous cue preference experi— ments that dealt with food selection and food versus sex. Since them, a preference has been demonstrated for vertical stripes over an inverted isosceles triangle (Berry, Rogers, and McCarrol, 1968). Vertical stripes were also preferred over horizontal stripes when rats were initially trained on a white-black discrimination task with white positive; when blagP_was positive, they subsequently preferred horizontal stripes. The reasons for such choice reversals are unknown (Law, 1954). Choice of stripes over triangles may have been based on relative amounts of contour present. Greater amounts of contour were preferred for checkerboard and random patterns, provided the internal elements were not too small (Karmel, 1969). 38 Rodent discrimination learning studies dealing with properties used to recognize objects could also provide some hints as to Peromyscus Visual cue preferences for orienting. Gibson (1969) Stated that "few studies with nonhuman animals exist on the prOperties of objects which serve as distinctive features." However, some such studies are available for the rat. Sutherland (1961) reviewed visual cue properties which rats use in discrimination learning. These prOperties include: .1) the ratio of contour to square root of area, 2) the ratio of horizontal to vertical axes (the importance of these factors is also suggested by the work of Dodwell and Niemi, 1967), 3) relative brightness relations of figure and background, and 4) shape of the lower half of the figure. Some of the studiespreviously reviewed may imply visual cuesgto use in a study of Peromyscus short distance orienta— tion. Presentation of vertical and horizontal cues to Peromyscus is suggested by rat spontaneous cue preference studies (Law, 1954; Berry, Rogers and McCarrol, 1968); by rat discrimination learning studies (i.e., the importance of the ratio of horizontal to vertical axes in shape discrimi— nation); by a Peromyscus artificial—habitat selection study with artificial trees (Harris, 1952); and by swimming Peromyscus orienting to the shoreline (Sheppe, 1965a). Presentation of a luminous cue is suggested by the sun compass orienting of Apodemus (Lfiters and Birukow, 1963), running wheel orientation of Peromyscus to an artificial "moon" ”-7 39 (Kavanau, 1968), and the importance of relative brightness relations of figure and background in rat discrimination learning studies. A topographical cue (relatively large in size but with low background contrast) is suggested by rat discrimination learning studies (particularly the importance of relative brightness relations of figure and background); by a running wheel study of Peromyscus which, according to the author, implies use of landmarks in orienting (Kavanau, 1968); and by the prevalence of such a cue in nearby wood- lots (personal observation). Conclusions on Home Range Orientation Except for rats, few studies have dealt with home range (short distance) orientation. Such studies have investigated the relative roles of the sensory modalities or have demon- strated the capability to use a particular modality. In spite of methodological problems, these studies imply that vision plays a major role in nocturnal rodent home range orientation. Scant information exists, however, for a noc— turnal rodent's visual cue preferences during home range orienting. Although Peromyscus clearly exhibits home range orienta- tion (as revealed by various incidental field observations), there are no studies on the modalities or cues Peromyscus uses in home range orienting. Indirect evidence suggests that any one or more of the modalities might be used, with vision being particularly important for longer distances. 40 For this reason, as well as others, my research concen— trated on Peromyscus visual cue orienting preferences. Visual cues to use in this study were selected utilizing information from rodent spontaneous cue preference and dis- crimination learning studies and having the cues abstractly represent natural objects. These procedures should maximize the applicability of this laboratory study's results to the natural home range orientation. General experimental features which may be desirable in a study of Peromyscus visual cue preference are as follows: 1) Young mice, circa 60 days of age, should be trained (Meier, 1964; Oliveria and Bovet, 1966). 2) Punishment should not be used in an orientation training procedure, for it may induce immobility and lack of choice responses (Bovet, 1965b). 3) Cues should be equidistant from the goal (Meyer, Treichler, and Meyer, 1965). PILOT STUDIES First Pilot Study The.first pilot study was an apparatus and training feasibility study; its methods differed considerably from all subsequent experimental series. With the aid of visual cues, a mouse could orient to one of eight holes along a circular arena, and then could enter a narrow L-shaped tube which led to a plastic box outside the arena. All eight boxes had sunflower seed rewards, but only the cue—associated box could be entered through an unlocked one-way door. After the mouse entered the box, I carried the entrapped mouse to a central release cylinder in preparation for the next trial. Although the mice showed signs of learning this task, the methods proved unsatisfactory: 1) they were too time-consuming, and 2) quite often the mice were obviously disturbed by the human intervention required (i.e., box transport and center releases). Consequently, the circular arena design was modified to the circular form described in detail in the definitive series methods section. Briefly, this consisted of eight clear plastic boxes, each with a lever inside, along the 41 42 arena's periphery. A tunnel led from the arena's center to the home cage below. When a trained water—deprived mouse pressed the cues-associated lever, it was rewarded with a drop of water in its home cage. Second Pilot Study In the second pilot study, deermouse orientation pref— erence for a visual versus a tactile cue was tested. Vision and touch were investigated for the following reasons: 1) In nature, cues of these two modalities would be rela— tively permanent for continual orientation use. 2) Various deermouse tracking studies have implicated vision and touch in orienting (my preliminary snow-tracking studies; Beer, 1961; Blair, 1951; Thomsen, 1945; and Sheppe, 1965b). 3) Vision and touch have been implicated in behaviors per-_ haps related to orientation, namely: habitat selection (Harris, 1952; Wecker, 1963), arboreal locomotion (Horner, 1954), and general locomotor activity as measured by activ— ity wheels (Kavanau, 1968). 4) These two modalities were amenable to careful experimental manipulation in my orienta— tion apparatus. The initial hypothesis was that woodland and prairie deermice would differ in their relative orientation prefer- ences for a visual and a tactile cue. Woodland deermice were expected to prefer the visual cue because they inhabit a visually Open environment, rely more on vision for 43 locomotion, and havelarger eyes. Prairie deermice were expected to prefer the tactile cue because they inhabit a closed visual environment of dense forbs and grasses, may use runways, and have smaller eyes. Three P, m, gracilis and two P, m, bairdi were tested. All were laboratory—reared, and-some were from lineages removed by more than four generations from the wild-caught ancestors. At the start of training the subjects were 103— l49-days old; at the end of testing they were 124-191 days old. Their apparatus and training stages were similar, but not identical, to those described in the methods section. Mice were tested for cue preference when they made no errors on 70% of a day's twenty—four (or more) trials on an eight—box training stage. For these routine (training) trials, two cues were associated with one box and were ro— tated to a different box after each trial. An "error" was scored for each lever a mouse pressed before it pressed the cues-associated lever. On test days, a mouse received 16-28 trials per day with routine training trials inserted between two types of test trials: 1) ”opposition" trials (the two cues were associated with boxes 180o apart from each other); and 2) "remoVal" trials (only one cue was present in the arena). Two different groups of mice were sequentially trained and tested. Mice in the first group were trained to go to. the "correct" box by using visual and tactile cues associated 44 with the neighboring 45o counter—clockwise box. Presumably, mice in nature would at times lack a prominent object near their goal, so they would have to maintain a fixed non-zero degree angle relative to some distant object, similar to a compass orientation. This arena cue set tested for such a postulated capability. The visual cue was a 14 cm long x 10 cm diameter cylinder with black and white vertical stripes 2 cm wide, mounted on the tOp edge of the metal arena, with a 37 cm x 10 cm black strip hanging downward from the cylinder's lower edge. The tactile cue was a strip of coarse sandpaper (52 cm x 20_cm) painted gray, extending from the tunnel exit to the box entrance. This strip was replaced every several days to eliminate accumulated odors. Mice in the second group were trained to go to the box directly in line with the cues (i.e., 00 angle orientation). The visual cue was the previously described cylinder with downward hanging strip. The tactile cue was a gray 50 cm x 10 cm coarse sandpaper strip with one end in contact with the "correct" box and the other end inserted under a 23 cm diameter gray sandpaper circle which fit around the tunnel exit. No matter at what point the mouse left the tunnel, it would make contact with the sandpaper circle and could follow its edge until it reached its junction with the long strip. Thus, the mouse could perceive both the visual and the tactile cue regardless of which way it faced when it entered the arena. Each mouse had its own long strip to re- duce any bias due to a prior mouse's odor. 45 Both species preferred the visual cue (Table 1). On Opposition trials the mice chose the visual cue more fre— quently; on removal trials, they made fewer errors when the visual cue was present than when the tactile cue was present. Although these data represent several days of testing, the same type of preference was usually shown each day. The initial hypothesis that prairie deermice would dif- fer from woodland deermice in the modality (tactile or visual) used for orienting was not supported. To conclusively reject such an hypothesis would require many cue sets, sys- tematically varying numerous stimulus parameters, and using larger numbers of subjects. Such a procedure would be un— attractive unless this pilot study suggested that such an ecological—modalities-difference hypothesis would be sup- ported. The hypothesis that deermice can maintain a fixed non— zero degree angle relative to some "distant" object was supported by the removal trial data when only the 450 visual cue remained. A general ability to utilize prominent Objects not located close to the mouse's goal could increase flexi— bility in a mouse's orienting tactics. It could also imply that mice may preferentially orient to certain objects even when they are not near the goal. Third Pilot Study The third pilot study used only visual cues because: 1) the previous series implicated vision rather than touch 46 oauo “mus muo “Hue one “one sno “mus eauo “sue mmcmu can mno “on> mno nou> muo uon> ouo kou> «no “ou> muouuo assume "mamflHD Hm>oemm me me me mm mm Amamaup coupsmommoc z OOOOfiO mm? Ram saw see sea saw one Hmsme> roars Go mamaHD coauflmommo x ..o x0 K. x0 a. maaflumum HOHHMQ Hpnflmn maaflumnm mfiaaomum Emma 00 xmmu me: . . mgnmeHDUHGME mDUmNEOHOm mo mmflommmnnm 03» ca OUCOHOMOHQ 050 ABV OHfluomu mdmno> A>v Hormw> .H OHQMB 47 in deermouse orientation, and 2) deermice have good visual acuity at short distances (Vestal, 1970) and thus could readily use visual cues for orienting. The initial hypothesis was that white-footed mice (P, leucopus) and prairie deermice (P, maniculatus bairdi) would differ in their relative orientation preferences for visual cues. For example, white—footed mice were expected to prefer an elevated object because they inhabit a visually Open environment, with prominent tree limbs and tall trunks, which they readily climb. Prairie deermice were expected to prefer ground level placement of the same Object because of their closed visual environment, relative lack of prominent visual objects overhead, and poorer climbing abilities (Horner, 1954). Six P, leuCOpus noveboracensis and five P, m, bairdi were tested on one or more cue sets. At the start of train- ing they were 57-77 days old; at the end of testing they were 133—261 days old. The mice were trained and tested in apparatus identical to that described for the methods sec- tion. For the first four cue sets, the mice had the training stages described in the methods section plus an eight box open stage with the cues rotated eagh trial. When a mouse made zero errors on at least 75% of this stage's 16 daily trials, it was advanced to the testing stage. The mice were trained and tested as two groups, each with a different cue set sequence. 48 Each mouse was tested for two days with twenty trials per day. The test trial sequence was identical to the second pilot study sequence (i.e., removal and opposition trials alternated with routine trials). Control trials were present at the end. During control trials the light was drastically dimmed as a check on whether the cues were really being used as visual cues. The mice were using the cues as visual cues. Control trials had an overall median of four errors, and a range of one of seven for the various individual mouse—cue set medians. Corresponding values for the immediately preceding routine trials were zero errors and a range of 0 to 0.5. These differences were statistically significant for each cue set (p g .05, one tail, Mann Whitney U Test). No consistent sex differences were present for opposi— tion trial cue preferences. For P. m. bairdi both sexes chose the same cue for 57% of the cue sets (N = 7); one sex had "no choice” for 14%; and the two sexes had different preferences for 29%. For P. leucopus cue sets (N = 5), the corresponding values were 60%, 40%, and 0%. The species variable in cue sets l—15 was unimportant for opposition trial choice. Both species chose the same cue on 70% of the cue sets, with similar choice magnitudes; both chose the same cue on 12%, but with considerable differ- ences in choice magnitude; one species had "no choice” on 18%; and the two species had different choices on 0%. 41.1 49 Direct Observation through a one—way vision screen (over a five-day period) also suggested that the species were similar in their motor orienting tactics. For runs from the tunnel to the arena periphery, both species had frequency distributions of "straight line run" > "tight looping run" > "wide looping run." For runs from one box to another, both species had "periphery paralleled run" > a run including the arena center > "shortest path" (straight—line run). Since sex and species differences were usually insig— nificant, overall cue choice will now be considered. Cues used in this pilot study are shown in Figure l. The cues ‘were 1) black (unless otherwise noted), 2) as odor free as -possib1e, 3) three dimensional, 4) lead weighted inside, 5) suspended unobtrusively, and 6) often analOgs of natural Objects. Various cue locations are shown in Figure 2. Cue counterbalancing procedures, identical to those described for the definitive series, were used. Choice results for individual cue sets are presented in Table 2; they yield considerable information on temporal and spatial parameters. When an item was shared between two one sets, choices on one cue set might bias choices on the next cue set (compare set la with 2b and 12 for prior exposure effects). The mice's visual scan angle includes points up to 450 above ground level but usually not those at 70°. (See sets 11, 12, la, lb and 2a. For the latter two sets' removal "£1 I'“ LG {3 H (D O’UOZEIT‘NCIHIEO'IIEIUOUIIV 50 Cues used in the third pilot study. black black black black ~b1ack black square square near bottom Of gray rectangle square near top of gray rectangle rectangle rectangle on gray rectangle stripes on gray rectangle two black circles on gray rectangle -black circle luminous circle white black black black black black circle vertical tube vertical tube vertical tube horizontal tube (X axis) horizontal tube (Y axis) gray bowl black which as a ' irregular figures on transparent plastic completely covered the top of the arena 'canopy" 51 K L M Q loorrect box /" J‘"\ Cues A-P, scale 1:5 Cue Q, scale 1:20 Figure 1 Figure 2. 52 Cue locations in the arena for the third pilot study. On the 1 _ 2... 3.. 4 _ 5— 6.. arena floor: near the arena center and directly in line with the correct box 450 counterclockwise from the correct box: next to one side of the neighboring box 45° counterclockwise from the correct box: blocking the entry hole of the neighboring box 180 counterclockwise from the correct box near the correct box and directly in line with it in juxtaposition with the correct box Above the arena floor: 7.. 8.. 9... 10 11 - on tOp of the correct box above the correct box further above the correct box, about halfway up the arena wall near the top of the arena at the top of the arena 53 -—-correct box .—wall —floor Scale 1:10 Figure 2 54 omlmm om mmaouau N .m> OHmGMDOOH ow .m> mm a mm "a was» Hmuaoufluon omnmm mom um maxmnx .m> Hmouuuum .z .m> Hm m solos was xon + m>onm ome can 00 . are .m> .a m ooanooa Oooa axon + m>onm Oon can Omc ~H< .m> one A as u 2 .mm pmmv mUOOnQdm msoum =Q= mmlmm pom mmmwuum .m> mamcmuomu mm .m> on a 0m "A ooauom mam "m Hoccsp Eonm How .m> How: am .m> HM m calms Ohm Axon + O>Onm 00b can ome use..m> mam N II I H O HIII om om me O.onn + O>Onm Om¢ can 00 o m m> d a d» n 2 MM Konuv muomflnsm msondt ..m_. X X Amuse :3 weaoso cofinfiuummc meanm mocoo umm Omsmm -Gmflpoz , .mcsem DOHAQ pnflflu on» as mHMHHD cowuflwommo How mmOHOQO Odo mdom>Eoumm .N OHQMB 55 .oa coaumooa O>Onm wapuouflc EU ONH mmB DH .N musmflm ca G3O£m Doc ma NH GOHDMOOQ£ .Umummg OHOB mOHmE BOG 038m .m msmOOOOH BOO m nufl3 Omomammu “post m osmoosoqm .Oovcommumou On our mOOaOHOmmap moaoomm acoumflmcoo QDAB mvwm O50 Mano .msmoosoamm "a “schema aaam "mm 6 U Q .N cam H mmndmflm comm .Ooscauaoomfip mmB OmOOE Odo .HMflHD OO>Hm o How xon =DOOHHOO= OS» we =xon + = .OOGHHHOOCO me 050 OOHHOMOHO OOHImm mmlmm OOHIom mwlmm mmlmm OOHImh whiom OOHIOm QOHIOQ mm "a DCOE mooa um Imomammflc HmOfiuHO> .m>_MMflmmmm mm .m> ddw. ma or end» HmOflDHm> .w> OHOHHO mOOGHEOH ma .m> dAM NH em H309 mmnm .m> OHOHHO msocflesa mm .m>.ddm Ha mm Dcoemomammwc “masmcm .m> Amwmmm HM .m> Hm OH or =mmocmo: .m> OHOHAO mOOGHEOH sad .m> dam a va H309 mmnm .m>-OQOD HMOMDHO> mm .m> dm. m mm "A om "m OHOHHU msocflfisa .m> MMHMN mH .m> dd h mu OHOHHO mOOGflEOH .m> MMMflm mH .m> dfl. o ooa "A OQOD HODGONHHOQ mm "m wanuN .m> flmmeMMN mo .m> Mm. m A@ n ZV :Q: cam :m: mQSOMmIEOHm muomflnsm Odom 56 trials with the 450 cue remaining, the overall median was zero errors, with individual medians ranging from 0 to 4.5; with the 700 cue remaining, the overall median was 5.3, with a range of 1 to 8.5). Deermice may have a stable hierarchy of choices for various spatial locations (see sets 12, 3a, 10 and 13). For objects close to the correct box, pattern— ing details were unimportant (see sets 4a and 4b). Table 2 also yields considerable information about ob— ject preference. A three—dimensional vertical Object was usually chosen over some other object (see sets 3b, 5, 8, and 12). P. leucopus seemed to show a stronger preference for such vertical objects than did P. maniculatus bairdi (see sets 3b and 5). Neither species chose cues that cast ground level shadows (see sets 5 and 9), nor did they choose luminous cues that were paired against similar nonluminous cues (see sets 6 and 7). However, when a luminous cue was paired against a dissimilar cue, the luminous cue was chosen (see sets 9 and 11). In conclusion, the initial hypothesis that prairie deermice differ from woodland deermice in the visual cues used for orienting was supported only on a few cue sets, and then the differences were unimpressive. Sex differences were also insignificant. Prior exposure effects may have dis— torted some of the cue set choices. However, all mice received the cue sets in the same order to facilitate the design of subsequent cue sets. Perhaps in nature experiential 57 factors also play a role in deermouse orientation, thus increasing the flexibility of orientation tactics. A pref- erence hierarchy for spatial locations may exist, but this inference is based on comparing cue sets having different space locations and objects. Therefore, the experimental study involved an investigation of the relative perferences of whitefooted mice for various objects and the spatial loca— tion of these objects in anticipation that a hierarchical order can be established among them. METHODS AND MATERIALS This series tested the visual cue preferences Of only P. leucopus males because 1) species and sex variables were generally insignificant in the third pilot study; 2) P. leucopus exhibited stronger choices than P. maniculatus on several cue sets that had analogs to natural objects; 3) males lack an estrous cycle, a potential source of varia— bility within an individual; and 4) concentration on one species and one sex permitted an in—depth analysis of orien- tation tactics. Three aspects of P. leucopus orientation behavior were examined: visual object preference, spatial location pref— erence, and the interaction of these two preferences. The general hypotheses for this series were as follows: 1) P. leucopus would prefer vertical objects. This hypothesis was suggested by snow—tracking and by laboratory pilot studies. 2) P. leucopus would prefer objects as close as possible to the goal. Research reviewed by Meyer, Treichler, and Meyer (1965) and my pilot study results suggested this hypothesis. 3) A hierarchy of relative orientation impor- tance would emerge for the three space parameters (angle from goal, radial distance from goal, and vertical distance from goal). My pilot study results implied this hypothesis. 58 59 4) When object preference is experimentally Opposed to space location preference, one of these two preferences would consistently take precedence over the other in all cue sets and by all mice. Relatively consistent Object and space preferences in the pilot studies suggested this hypothesis. subjects Six P, leucopus noveboracensis males were each tested on nine cue sets, with a different cue set sequence for each mouse. The mice were genetically heterogeneous and lacked experience with the natural habitat; they were all from different laboratory-reared parents (which were at most two generations removed from the wild-caught ancestors). At the start of training, the subjects were 55-73 days Old; at the end of testing they were 169—197 days old. Prior to training, each male was housed for about three weeks with a similarly aged female (usually its sibling) in a large plastic cage, 49 cm x 38 cm x 20 cm. This cage con— tained wood chips, cotton nesting material, a floor level food container, a water bottle, and an activity wheel. Each male thus had some early experience in orienting to animate and inanimate Objects. Such experience might facilitate orientation in an arena. Mice were maintained, trained, and tested on a reverse light cycle. During the 8 a.m. to 8 p.m. "dark" phase, only a 7.5 watt white bulb was on; during the "light" phase, a 100 watt white bulb was also on. 60 Throughout training the mice were individually housed in laboratory plastic cages, 28.6 cm x 12.7 cm‘x 15.2 cm, provided with wood shavings and cotton bedding. Cages were kept 3 to 8 cm apart on shelves located between the test room and the training room. The mice were motivated by controlling their drinking schedules. On weekdays the subjects Obtained water only during the training or testing session, which occurred approximately the same time each day for a given individual. (They Obtained about 0.6 ml of water during the orientation task, and a 0.6 gm apple chunk after completion of the task.) Subjects maintained an overall median body weight of 90% of initial body weight (defined as the body weight attained on the day prior to the start of training). On the weekends the subjects were allowed ad libitum water for about 24 hours. Throughout the week, food was present ad libitum except during orientation tasks lasting less than five hours. Apparatus Early Training Stages The mice were initially trained to press a lever and to obtain water from a fountain. During training, a mouse's cage fitted flush against a training module. Four mice could be trained simultaneously, each with a separate plastic module (33.0 cm x 0.5 cm x 15.9 cm) having two 3.2 cm holes. One hole could accommodate any one of the following objects 61 associated with lever—pressing: a lever protruding into the cage, or a plastic box with a lever inside, or a tunnel which led to a lever. The other hole led to a water foun— tain assembly (Figure 3). A mouse had to exert a 3 gm force to activate the lever, a Robertshaw microswitch, mounted on a wooden support frame. The lever protruded about 3 cm into either the mouse's cage or the plastic box. Wire—reinforced corrugated 5.1 cm di— ameter tubes could be connected to two of the modules. These 63.5 cm long tunnels each exited at the center of an arena 55.9 cm in diameter and 55.9 cm high (Figure 3). The inner walls and the masonite floor of the arena were painted a flat medium gray. A plastic box with lever inside was present along the arena's periphery. Each lever operated an Esterline-Angus event recorder channel, a water-delivery Skinner solenoid valve, and (in the last training stage) a muted door bell. When amouse pressed a lever, a 0.03 ml drop of water appeared on the fountain: an activated solenoid valve allowed water to flow from a reservoir funnel, through plastic tubing, to a hollow brass rod fountain. Large Arena Stages The mice received their final training stages and were tested for cue preference in a large arena, 116.5 cm in diameter and 56.8 cm high (Figures 2 and 4). A mouse's cage fitted flush against a 30.5 cm x 15.9 cm x 0.3 cm Opaque 62 macho chcsu hanewmmm aflmucsom HODMB hupcm HOCGOU OHOUOE ofiu mo xomn CH x09 OHOOOE any EOHM mawpsuuonm HO>OH OHOUOE mcHCHMHu .mwmmum oneness» mHHmw CH cow: msumummmd .m «munmmo musmflm 63 m OHSOHM .I. I. —I 7 l / . , L O arms. 0 a. o Wfl O mo. m m \ m _ m a _ .m \\\\\\\\IIM\ . \\\\\\\II KVIIIIl/llll 1_.false ceiling \ VII ‘ ’ ’ ~: :4; '51; I . I i -” I I I I. 'v ”I, n I ‘ I, “I I l‘ I I I! I“ '\ . "I I. II I \‘ A“ . I ; I . ’ I I II I :1, A .1 ; . | v . {1' : " I, I I‘ 1‘ I: Iir': ’1 ’ ' _. " . ,1 I: I I:- ' I . . ’II 'I - a... ”Swat I I .1'1' . . I _————— curtain I arena wall arena floor tunnel fountain guillotine door ,,,.a-"' entry hole cage Figure 4. Front view of large arena. 65 plastic module, located 27.9 cm below the large arena's floor. This module had a tunnel entry hole, an opaque guillotine door capable of blocking the hole, and a "water fountain" assembly with a "drinkometer." An opaque card— board light shield surrounded the mouse's cage. A mouse could travel between its cage and the arena's center through a 63.8 cm long x 5.1 cm diameter tunnel. When necessary, I could detain a mouse in its cage by block— ing the tunnel entrance. A guillotine door was lowered from my station outside the arena room via a noiseless string and glass tube pulley system. When a mouse lapped at the foun— tain, it activated a "drinkometer" which consisted of an electrified 6 cm x 7 cm hardware—cloth platform and fountain tube, both connected to a Sigma relay with coil potentiometer and transistor. The relay was connected to an event recorder channel. Basically, this was equivalent to V. J. Polidora's drinkometer design (personal communication to J. A. King, 1965). The arena and surroundings within the mouse's visual field were rendered a homogeneous gray. This was expected to facilitate the orientation learning task by eliminating most extraneous visual cues. The masonite floor and aluminum walls of the arena were painted a medium flat gray. A medium gray cloth extended from the top of the arena wall to a false ceiling made of the same material (Figure 3). The mouse could enter eight clear plastic boxes, each 12.7 cm x 6.7 cm x 6.4 cm, mounted 40.5 cm apart along the 66 arena's periphery. Each box had a 3.2 cm entry hole and a slotted plastic track that could hold a plastic door. During the first three training stages plastic doors blocked the entrances to some of the boxes. In all training stages and in the test stage, a mouse could obtain water only by pressing a lever in a cue- associated ("correct") box. Asgangswitch.setting allowed only the "correct" lever to activate a solenoid water delivery valve and a muted doorbell, which signalled a water reward. The levers of two Robertshaw microswitches were soldered together and protruded into a box. This tandem arrangement and a toggleswitch-gangswitch combination allowed a mouse to obtain water by pressing the lever in either of the two cue—associated boxes during opposition trials; by preventing simultaneous firing of two event recorder pens, it also allowed me to know which lever the mouse had pressed. One microswitch lever in a box connected with one of eight event recorder channels located outside the arena room. The tan— dem microswitch (and its opposition trial counterpart 180O away) connected to one of four toggle switches. During opposition trials one switch was turned on, allowing water delivery by lever-pressing in either of the two opposed boxes. Light levels in the large arena room approximated levels mice would encounter in the natural habitat. A ceiling 7.5 watt white bulb with a ground glass defractor yielded 0.045 f.c. at the arena center and 0.015 f.c. at the arena 67 periphery. The value at the periphery was equivalent to nocturnal full moon illumination (Kavanau, 1967). During control runs less than 0.0001 f.c. illuminated the arena center. The only light source was a 7.5 watt white night light mounted below the arena center and covered by an opaque box with a small opening (0.8 cm x 0.8 cm). A monitoring station was located outside the large arena room. Near the event recorder were a stOpwatch and a knife switch. These permitted me to obtain lever pressing sequences and latencies and to independently operate the water delivery system and bell. When necessary, I could quickly lure a mouse into its home cage for transport to its maintenance room. Procedures Early Training Stages Early training stages were necessary because the large arena's orientation task was complex: after a mouse pressed the correct lever, it had to run away from the lever and re— turn to the home cage for water. Thus, both spatial and time lags were present for response reinforcement. During transfer to the training module, certain.pro- cedures prevented directly handling the mice and the possi- bility of escape. The lid of a mouse's cage was removed by sliding under it a 28.6 cm x 18.4 cm x 0.3 cm clear plastic piece. Then the cage was fitted against the module and the plastic piece was slid away. 68 To obtain water, a thirsty mouse learned to press a lever-and then go to the water fountain. In the first training stage, the mouse pressed a lever protruding into its cage; in the second stage, it entered a small plastic box and pressed the enclosed lever. Unlike the first stage, second stage lever-pressing could not be due to accidental stereotyped jumping. In the third stage, a mouse ran up a tunnel to a small arena's center, ran to the peripheral location of a single small plastic box, entered the box and pressed the enclosed lever; then it had to run down the tunnel to obtain water from the module fountain. This third stage task was similar to the large arena orientation tasks. The mice were initially alerted each day to the salient features of the training stages by seed baiting and my lever pressing. For example, in the first stage, sunflower seeds were placed near the lever and the water fountain. I pressed the lever several times in rapid succession; this resulted in a "free" drop of water on the fountain. Intermittently, I directly observed the mice. The mice were left in the training apparatus no more than 36 hours for the first training session of any stage; they could obtain water only by pressing a lever. On subse- quent (daily) sessions they usually had no more than two hours to obtain water. Mice were advanced to the next train— ing Stage when they pressed a lever at least 16 times in two hours or less. They readily learned these tasks, for 87% of 69 all g. leucoEus (N = 23) could be trained on all three stages, with a median performance of one day to reach the advancement criteria, and a range of one to three days. Large Arena Stages A mouse's basic orientation task was to run up the tunnel from its cage and press the cues—associated ("correct") lever in the arena. A muted door bell provided an immediate secondary reinforcement: it indicated correctness of re— sponse and the presence of water on the module's fountain. The mouse then ran down the tunnel and over to the fountain. Its lapping activated a "drinkometer" and event recorder pen. When I knew that the mouse was not in its cage, I could de- tain it by lowering a guillotine door. Then, I could enter the large arena room and, unobserved by the mouse, place cues by another box selected, with the aid of "curtailed" randomization techniques (Cox, 1958). A trained mouse was minimally influenced by my activi— ties. After placement of the cues, I deliberately tapped each 900 arena sector to minimize the mouse's use of cue placement noises for orienting. Then I returned to the moni— toring station and raised the guillotine door. By having a guillotine door to momentarily trap a mouse in its home cage and by having monitoring activities outside the arena room, there was minimal disturbance to the mouse. Several procedures kept an orienting mouse dark-adapted. It was transferred to the large arena room under red light 70 (7.5 watts) or with an opaque light shield over its cage. In the large arena room, the large opaque cage shield main— tained low light levels for the caged mouse at all times. Training stages of gradually increasing difficulty pre— vented performance "extinction" from occurring. For the first cue set only one box was open for entry; then, two; four; eight (with one box "correct" throughout a daily session); eight (One box "correct” twice in succession and all eight boxes "correct" sometime during the daily session); and eight (testing). For all subsequent cue sets, the mouse was started on the first eight box training stage. On all stages except the first stage and the first eight box stage, the cues were moved to another box after the mouse had received water several times ("trials") on the previous box. In the training stages the major dependent variable was the number of "errors"——presses on levers that did not have the visual cues associated with them (i.e., "incorrect" presses). Here, each individually distinguishable event recorder blip was scored as an "error." Mice were advanced from one training stage to the next when at least 16 "correct" lever presses were made in an hour, with one or more boxes having a ratio of three "correct" presses for every two "incorrect" presses. During the last eight box training stage, a mouse was given daily terminal practice trials for boxes with many errors. For advancement to the testing stage, a ratio of at least two I'correct" presses for every one ”incorrect" press 71 was required. This criterion of no more than 8 errors in 16 trials is analogous to a two—choice 90% learning criterion. (Both criteria yield binomial probabilities for obtaining the performance level by chance alone of about 0.002.) However, these criteria were not necessarily applied to the first successive 16 trials. Intermittent loud noises or strong odors from various sources in the building seemed to momentarily disturb and disorient the mice. Consequently, I discarded as many as six such "disturbance" trials from the data and re—presented the boxes in question later in the training session. This procedure was necessary for almost 60% of all mouse—cue set combinations. To obtain six mice ready for testing, I had to train eleven mice on the large arena stages: only 55% of the mice were able to reach learning criteria in all of the large arena training stages. The six subjects actually tested had a median of 1 day (with a range of 193 days) to reach advance— ment criteria during the one, two, and four box training stages. They had a median of 1 day (with a range of l—lO days) for the first eight box stage of the various cue sets, and a median of 2 days (with a range of lrl7 days) for the second (last) eight box training stage. Frequently, the mice performed at better than criterion level on the final day of this last training stage; their median number of total errors was 6, and on 40% of all mouse—cue set combinations their performance was in a range of 0—4 total errors. 72 Testing, which began the next day, consisted of "routine“, "removal", "opposition", and "control” trials. To reduce the possibility of a mouse's orienting to the box previously correct, on the next trial I placed the cues at least 900 away from their previous location. On routine trials, both cues were associated with one box. These trials were interspersed between the other test trials in order to maintain a high performance level (i.e., few errors) and to lessen potential carry—over (biasing) effects of one test trial on another. During the removal trials, one of the two cues was re- moved, and the mouse could orient only to the remaining cue. If the mouse still oriented efficiently (i.e., if its median errors approached zero), the remaining cue was preferred. Conversely, if the mouse became disoriented, this cue was unpreferred. Both cues were present in the large arena during the opposition trials, but the two cues were near different boxes, which were 1800 apart. The two boxes, equidistant from the correct box on the preceding (routine) trial, were both correct in terms of the water reward. Thus, during opposition trials the mouse had a simultaneous choice situa— tion. The major dependent variable was the number of trials on which the mouse chose a particular cue. Control trials were designed to test for possible orien— tation to nonvisual aspects of the experimental cues. Both cues were near one box, but the light was greatly reduced so 73 that a mouse could not see the cues. If the mouse had really been using the cues as visual cues, it should now be dis— oriented. Conversely, if it had oriented to auditory, olfactory, or tactile aspects of the cues, then temporary removal of the visual sense should not result in an increase in the number of errors. For each cue set, there were two days of testing with 24 trials per day. The test trial sequence was as follows: routine — removal — removal; routine — opposition...routine — opposition (trial 19); routine (trial 20) - control — routine — routine — control. (Two control trials similarly appeared on the day just prior to testing.) Thus, on each test day there were 12 routine, 2 removal, 8 opposition, and 2 control trials. The test day sequence of correct boxes, assigned before I tested a mouse, was designed to minimize certain potential biases: a mouse might choose a box during opposition trials on the basis of two different types of box location prefer— ences rather than on the basis of cue preferences. For the first type of bias, a mouse might make a series of choices due to prior experience with an identical (or very similar) numerical sequence. I eliminated this bias by having differ— ent numerical sequences, namely: 1) For each mouse, a numerical sequence was as different as possible from the day's preceding sequence. 2) Each mouse received a particu— lar numerical sequence only page during the entire experiment. 3) For a given cue set, each of the mice received different 74 numerical sequences. In the second type of bias, a mouse might prefer a certain individual box, regardless of which opposition trial cue was associated with it. I controlled for-this bias by having every box correct an equal number of times during the "routine—opposition" segment (trials 4-19) of each test day. There were eight routine and eight opposi— tion trials in this segment. Thus, each box was correct for one of these routine trials and for two of the opposition trials (on one trial it was associated with one of the two cues; on the other trial, with the gghgg cue). During the daily removal test trials, each cue was present on one of the two trials. The temporal order for its presence was counterbalanced for the two test days and also was counterbalanced between mice. A test day's two removal trials were closely matched for relative difficulty (indi— cated by the number of errors associated with each box on the previous day). A control trial and its immediately preceding routine trial were also closely matched for relative difficulty on the basis of data from the mouse's previous trials of that day. Usually, a mouse received a different correct box for each control trial in a cue set. The nine cue sets were presented in sequential blocks ("phases") of three cue sets each (Table 3). In Phase I, object preferences were tested (both objects occupied similar spatial locations); in Phase II, spatial location preferences 75 Table 3. Cue sets used in the experimental study. _‘__ Object Lecation Cue shown in shown.in set Objects Figure l - Spatial locations Figure 2 Object Preference Phase 1 Black vertical K, N Both cues placed on 7 tube gs. (both top of the correct (both) black horizon— 12.4 cm x box. tal tube. 3.0 cm di— ameter) 2 Black vertical K Tube 10 cm in front 5 tube yg. of forward corner inverted gray P of correct box. bowl (12.4 cm Bowl in contact 6 x 23 cm base- with other forward 17 cm top corner of box and diameter). with arena wall. 3 Black vertical K Tube hung downward 10 tube yg. from arena's top (both) luminous I edge 38 cm above a circle (5.1 cm back corner of cor— diameter). rect box. Luminous circle, similarly placed, 44 cm above box and 13 cm away from tube. Spatial Location Phase 4 2 identical both Near the arena' 5 1 black vertical K cgnter, 34 cm and tubes. from the correct box vs. near the 3 periphery, 34 cm and 450 counterclockwise from the correct box, blocking the neighbor— ing box's hole (radial Kg. peripheral dis— placement). 5 2 identical both 9n the correct box 7 black vertical K yg. 38 cm directly 10 tubes. above it (a vertical displacement cue set). continued 76 Table~3——continued . Object Lecation Cue shown in shown in Set .Objects Figure 1 Spatial locations Figure 2 6 2 identical both 38 cm directly 10 black vertical K above the correct tubes box (93 angle) gs. near the periphery 3 (on the floor), 34 cm and 45 counter- clockwise from the correct box, block- in the neighboring box's hole (vertical versus‘peripheral displacement). Object Preferences vs. Spatial Preferences Phase 7 Vertical tube K Luminous I circle (Note: the tube at location 10. Tube blocked the 45° 3 counterclockwise neighboring bos's hole. Circle was 44 cm 10 directly above correct box. For two mice the circle was at location 3 and Circle was mounted on a small gray-fronted battery clip.) 8 Vertical tube K Gray bowl P (Note: Tube was near the l arena's center and 00 from correct box. Bowl blocked the 45° 3 counterclockwise neighboring box's hole. For one mouse the tube was to one side of loca- tion 10, and the gray bowl at location 6 was diagonally opposite the tube.) 9 Vertical tube K Vertical tube 3 blocked the 45° counterclockwise neighboring box's hole. Horizontal tube N Horizontal tube 7 lay on top of the correct box. 77 were tested (identical vertical objects occupied the various spatial locations); and in Phase III, the relative importance of object preference versus spatial preference was tested (a preferred Object in a non~preferred spatial location was tested against a non-preferred object in a preferred spatial location). Object preference was tested before spatial loca« tion preference to minimize carry—over (biasing) effects from the vertical object. To accurately test object versus space preferences, I assigned each mouse Phase III cue sets that reflected its gwg prior choices of objects and spatial locations. All three phases had several features in common. Identical black tubes were present in all nine cue sets and various objects were approximately equidistant from the cor— rect box in seven of the nine cue sets. For each phase, a "curtailed" randomization procedure was designed to reduce potential carry-over of cue preferences from one cue set to the next due to prior exposure. I randomly assigned the cue set order of presentation to each mouse. However, only 96 of 216 cue set combinations were available for assignment in order to reduce presentation of identical items in sequential cue sets. For example, the order (1) A vs. B, (2) B vs. C. and (3) C vs. D has B and C items presented sequentially and was omitted; in contrast, the order (3), (l), and (2) has just B presented sequentially and was acceptable. ,In the appropriate cue sets in all three phases, counter- balancing procedures were used for right—left locations of 78 cues; for the two vertical tubes' placement; and for the exact placement of horizontal versus vertical tubes on tOp of the two correct boxes during an Opposition trial (i.e., each was over a box's geometric center, or both were forward, or the vertical cue was back and the horizontal cue was forward). Statistical Analysis The major dependent variable, the frequency with which a mouse chose one of the two cues during opposition trials, was analyzed with two different statistical tests. For each cue set, a chi square 1 df test analyzed the combined results. which totaled 96 observations (16 Opposition trials per mouse x 6 mice). This test assessed the significance of the deviations from an expected one to one ratio of no choice (48 opposition trial choices of one cue and 48 of the other). To obtain a better fit to the theoreticale.2 distribution, the conservative practice of subtracting Yate's correction factor from theXfia numerator was followed. In order for the chi—square to be valid for the combined data, the individual opposition trials-had to be independent of each other and the mice had to be similar in choice behavior. A runs test for each mouse-cue set combination indicated that the indi- vidual opposition trials were indeed independent of each other. The assumption that the mice were similar in their choice behavior was checked by analyzing the same choice data with 79 a "t" test having 5 df. This test compared the overall mean number of trials per mouse an object was chosen with the theoretical "no choice" mean of 8 (16 x 50%). If the mice were similar in choice behavior and preferred one cue over the other, the standard error would be small and a statisti- cally significant difference would be obtained between observed and theoretical mean number of object choices. Various factors, which could potentially bias the re- sults of bothjZ? and t tests, were analyzed by bgth paired comparisons tests (Wilcoxon or Walsh tests) and an unpaired test (Mann Whitney U test) and were all statistically non— significant. These tests indicated that possible differences in choice behavior for the two test days, the two subject replicates, the two vertical tubes, and right-left locations of a cue did not bias the analysis of the opposition choice frequencies. Other factors were analyzed by Friedman tests. These factors involved possible differences in choice be~ havior for the temporal order in which a cue set appeared in a phase, and for the exact geometrical box-tOp placement of the horizontal and vertical tubes during Opposition trials. For the other measures of cue choice, errors and lat— encies, I used the following statistical tests: Wilcoxon or Walsh tests (N = 6) and_Mann Whitney U tests (n1 = n; = 6). Comparisons were used on medians because unequal sample sizes were frequently present and the data were sometimes skewed. Preliminary tests, identical to the tests described above, 80 analyzed for differences in a dependent variable across the two test days. These tests almost always yielded nonsig— nificant differences. For those few instances where test days were significantly different, separate analyses were done for each test day. Both Wilcoxon or Walsh tests (N = 6) and Mann Whitney U tests (n1 a n; = 6) were also used for comparisons of control trial medians with routine, removal, or Opposition trial medians. Two-tailed tests ( a = .05 ) were used for all results analyzed in this study unless other- wise indicated. RESULTS Cue preference data are presented for each of the three phases in order, namely: object preference, spatial loca- tion preference, and object preference versus spatial location preference phases. Evidence is presented that the mice preferred certain boxes, that they used the cues as visual cues, that they could clearly see each cue, and that they were well trained prior to testing. Object Preferences For object preference (Phase I), the mice significant~ ly preferred the vertical cue over any of the other cues presented, as revealed by opposition trial choices (Figure 5, and Table 4). Overall, there was a moderate degree of choice preference for vertical tube over horizontal tube (t: 2.99, df= 5, p < .05; X2: df= 1, p: .05) and a stronger degree of choice preference for vertical tube over gray bowl (t = 5.27, p < .01;)KFZ: p < .001) and perhaps a preference for vertical tube over luminous circle (t = 1.58 p < 0.20; butng p < .001). The luminous circle-vertical tube results were statis— tically ambiguous, for they yielded statistical significance 81 Figure 5. 82 Means and standard errors of object prefer- ences on opposition trials. Each histogram shows the number of choices for the underlined object. The dashed line indicates the theo— retical mean number of trials that would be obtained if the mice equally preferred each of the two cues. Trials 83 161 n————-| 8~-~ _._._._ 0_4 Horizontal Gray Bowl Tube vs. vs. Vertical Vertical Tube Tube Figure 5 Luminous Circle vs. YerJLisél Tube Table 4. Number of trials the underlined object was chosen by individual mice when objects were Opposed. ‘ Random choice would involve fluctuations above and below the 50% level (8 choices). Note: these values are consistently above this level. Individual mice No. No. No. No. No. No. l 2 3 4 5 6 Horizontal tube — vertical_tube 8 12 9 9 10 10 (Figure l, N vs. K) Gray bowl - Vertical tube ll 10 ll 14 10 11 (Figure l, P vs. K) Luminous circle - Vertical tube 15 15 12 15 2 9 (Figure 1, I—vs. K) 85 with the If: df test, but not with the t test. A Dixon Gap outlier test suggests that mouse number 5's data probably did not come from the same pOpulation as the other values (R1 = 0.54, critical R1 = 0.56, p = .05+). Since "bias from rejecting a valid observation is usually far less than that caused by retaining a contaminant" (Bliss, 1967, p. 154), mouse number 5's value was consequently discarded. Reanalysis of the data, with five mice, yielded statistical significance for both tests: t = 4.33, df = 4, p < .02; and for )0, p < .001. The error scores and removal trial latencies had low values (Table 5). None yielded statistically significant differences within the three cue sets (Wilcoxon or Walsh tests, and Mann Whitney U tests). Thus, only opposition cue choice frequency indicated any Cue preference. Spatial Location Preferences The Phase II spatial location preferences can be arranged in a hierarchy (Figure 6, and Table 6): 1) Radial displace- ment, the position of the cue along a radius of the arena, interfered m9§t_with a mouse's orientation. Here the prefer- red location was at the arena's periphery, even though a 450 angle orientation had to be maintained (t = 2.45, p < 0.10; )12: p = .02). 2) Peripheral displacement interfered to a lesser extent with a mouse's orientation. When both vertical tubes were located near the arena's periphery, the cue Table 5. . overall medians 86 Test errors and latenCies for object preferences: and ranges. -— Opposition Removal Removal errors errors latencies (seconds) Horizontal tube 0 O 29.0 (0—0) (0—1.5) (15.0-120.5) vs. Vertical tube 0 0.5 32.8 (0-0) (0-2.5) (22.0-62.0) N.S. N.S. N.S. Gray bowl 0 2.3 55.0 (0—1) (0-3.5) (13.04115.O) vs. Vertical tube 0 0.8 32.8 (0-0) (0—2.0) (14.0-127.0) N.S. N.S. N.S. Luminous circle 0 1.0 21.8 (0-1) (0—3.0) (10.5-137.5) vs. Vertical tube 0 0.8 52.3 (0—3.5) (0-5.0) (15.0—174.5) N.S. N. N.S. Figure 6. 87 Means and standard errors of cues displaced along various spatial coordinates on opposition trials. Each histogram shows the number of choices for the underlined.cue. The dashed line indicates the theoretical mean number of trials that would be obtained if the mice equally preferred each of the two displace— ments. Trials 88 16W 1 1 &~——- ----- l —-——-- Radial gg_box Peripheral vs. vs. vs. Peripheral Above box Vertical Figure 6 89 Table 6. Number of trials a particular displacement of a vertical tube (underlined) was chosen by indi- vidual mice when displacements were Opposed. Random choice would involve fluctuations above and below the 50% level (8 choices). Note: these values are consistently above this level for all but the last cue set listed here. Individual mice No. No. No. No. No. No. l 2 3 4 5 6 Radial displacement vs. Peripheral displacement 10 9 13 7 11 10 (Figure 2, 1 vs. 3) 93 above box vs. 45° above box (vertical 9 8 7 10 6 9 displacement) (Figure 2, 7 vs. 10) Peripheral displacement vs. Vertical displacement 12 16 13 5 11 11 (Figure 2, 3 vs. 10) 90 located 00 from the correct box was chosen (even though it was quite high above the box) in preference to the cue located 450 from the correct box, and at floor level (t = 2.26, p < 0.10; X 2: p < .001) . 3) Vertical displacement did g9; interfere with a mouse's orientation. Locations near floor level and about 450 above floor level were about equally preferred (t = 0.28, p ) 0.50;:K.2 : p >> .05). Two of the cue sets' results were statistically ambigu— ous, with t tests yielding borderline significance ( p < 0.10) and.}:: df tests yielding significance. A range/ standard deviation outlier test suggests that mouse number 4's data probably did not come from the same population as the other values for the radial versus angular displacement cue set (R/SD e 3.00, critical R/SD = 3.01, p e .05+). Analysis of the remaining data yielded statistical signifi- cance (t = 3.83, df = 4, p < .02; )(2: p < .01). Mouse number 4's data can also be omitted for the angular versus vertical displacement cue set (R/SD = 3.04, p < .05). Reanalysis yielded statistical significance (t = 4.96, df = 4, p < .01;)62: p < .001). The error scores and removal trial latencies had low values (Table 7). None yielded statistically significant differences within the three cue sets. 91 Table 7. Test errors and latencies for spatial displacement . preferences: overall medians and ranges. Opposition Removal Removal errors errors latencies (seconds) Radial displacement 0.3 2.5 36.3 (0-2) (0-9.0) (20.5-120.0) vs. Peripheral displacement 0.8 0.5 27.5 (0.2) (0-3.5) (15.0-42.5) (Figure 2, 1 vs. 3) N.S N.S. N.S. 0° above box 0 0.8 34.3 (0—0) (0—2.0) (14.0-136.5) vs. 45° above box 0 0.3 21.8 (0-0) (0-2.0) (17.5—86.5) (Figure 2, 7 vs. 10) N.S. N.S. N.S. Peripheral displacement 0 0.8 57.8 (0-1) (O—2.0) (12.5-136.5) vs. Vertical displacement 0 1.3 30.0 (0-0) (0—5.0) (16.5—194.0) (Figure 2, 3 vs. 10) N.S. N.S. N.S. 92 Object Preferencesyversus Spatial Location Preferences When object preference was tested against spatial loca— tion preference (Phase III), the mice oriented to the pre- ferred object on two of the three cue sets (Figure 7, and Table 8). They showed a statistically significant preference for the radially displaced vertical tube over a peripherally displaced gray bowl (t = 3.22, p < .05;)(2: p < .001), but not in the set which included a vertically displaced luminous circle (t = 1.14, p < .40;:X3: p > .05). They showed a statistically significant preference for the previously pre- ferred spatial location in the cue set which included a horizontal tube (t = 4.44, p‘< .Ol;;('2; p < .01). As in the prior phases, the error scores and removal trial latencies had low values (Table 9). Test Data Other than Cue Preference Data P. leuc0pus preferred not only certain cues but also certain boxes (Figure 8). Boxes 2, 3, and 4 were signifie cantly preferred over boxes 6, 7, and 8 ()12 goodness of fit test: )32 = 77.15, df = 7, p < .001). Analysis of the data for each individual mouse revealed similar preferences, with three mice showing statistically significant box preferences. Such box preferences can not bias the cue preferences, however, for each box was represented equally each day, and a different numerical presentation sequence was used for each test day and for each mouse. Figure 7. 93 Means and standard errors of Object versus spatial displacement preferences on Opposition trials. Each histogram shows the number of choices for the underlined cue. Captions for histograms are for the usual cue set arrangement. Each histogram represents the mean and standard error for all six mice. Trials 94 16- l 8!—— ---———-— —-—-—-—-—-——-———-—-——-—-—-— T OJ Luminous Gray bowl Horizontal Circle (Peripherally Tube (Vertically displaced) (On correct displaced) vs. box) vs. vs. Vertical Tube Vertical Tube Vertical Tube (Radially_ (Peripherally (Peripherally_ displaced) displaced) displaced) Figure 7 Table 8. 95 Number of trials the underlined cue set was chosen by individual mice when object preference was Op— posed to spatial displacement preference. Random choice would involve fluctuations above and below the 50% level (8 choices).- Note: these values are consistently above this level for two cue sets, and consistently below the level for the last cue set. Arrangements different from the indicated usual arrangement were a = gray bowl (next to the correct box) vs. vertical tube (above the correct box), and b = luminous circle (450 from the correct box) vs. vertical tube (above the correct box). Individual mice No. No. No. No. No. No. l 2 3 4 5 6 Luminous circle (vertically displaced) vs. b b Vertical tube (peripherally 10 10 13 ll 5 7 displaced) Gray bowl (peripherally displaced) vs. Vertical tube (radially 15 10 9 9 ll 12 displaced) Horizon tube (on correct box) vs. Vertical tube (peripherally 6 5 7 4 4 7 displaced) Table 9. 96 Test errors and latencies for object preferences versus spatial displacement preferences during Opposition trials. Opposition Removal Removal errors errors Vlatencies (seconds) Luminous circle 0 0.8 58.0 (vertically displaced) (0-1) (0—3.0) (13.5-144.5) vs. Vertical tube 0 0.8 34.0 (peripherally (0—1) (0-2.0) (ll.0_7l.5) displaced) N.S N.S p e .05 Gray bowl 0.3 2.3 41.3 (Peripherally (0—2) (0—4.5) (14.0—130) displaced) Vs. Vertical 0 1.5 34.0 (radially displaced) (O—l) (0.5—2.5) (21.0—125)' N.S. N.S. N.S. Horizontal tube 0 0.5 29.0 (on the correct box) (0—0) (0—l.5) (16.0—92.5) vs. Vertical tube 0 1.0 35.3 (peripherally (0-0.5) (0—5.0) (23.5-l7l.5) displaced) N.S N.S N.S. Figure 8. 97 Total number of opposition trials on which a box was chosen. The dashed line represents the ex— pected number of choices for each box if all boxes were equally preferred by the mice. Trials 150 ‘ 100 J 50 a 98 Box Number Figure 8 99 The mice chose their cues during the three phases by the visual characteristics of the cues. Control trials, with ambient light negligible, had significantly more errors than the immediately preceding routine trials for all cue sets (Table 10). Such a result would be expected with the temporary elimination of the visual sense, but would not be expected if the mice chose the cues on the basis of odor, touch, or echolocation. An alternative explanation, however, for the results in Table 10 is that the behavior of the mice (apart from vision) was affected by a drastic reduction in the room's light level, and consequently their nonvisual orientation performance was impaired. If this behavioral change hypother sis were true, the latency to the first lever press on the control trials might be significantly different from the immediately preceding routine trial's first-press latency. But (with the exception of one cue set), there were no sig— nificant differences between the two trials' first-press latencies (Table 11). Thus, the increased number of errors on control trials is attributable to a temporary elimination of the visual sense rather than to a change in the mice's general behavior. The mice used only the experimental visual cues pro- vided: they did not use extraneous cues (arena wall scratches, etc.) in orienting. The possible use of extraneous cues was examined by comparing actual and theoretical frequencies with 100 Table 10. Number of errors (medians and ranges) in sequential control and routine trials. Probability Cue set Control Routine (Mann Whitney test) 1 2.3 0 (0-5) (0-0) .02 2 3.5 0 (0-5) (0-0) .02 3 5.0 0.3 (2.5—7.5) (0—1) .002 4 1.5 0.5 (1-395) (0‘1) .02 5 3.0 0 (1.5—3) (0-0) °02 6 1.5 0 (0.5—4) (0-0) .002 7 0.8 O (O-l.5) (O—O.5) N.S 8 2.0 0 (1—5) (0-2) .02 9 1.3 0 (0—2) (0-0) .02 101 Table 11. Latencies in seconds (medians and ranges) for the first lever pressed in sequential control and routine trials. Probability Cue set Control Routine (Mann Whitney test) 1 30.5 15.0 (l9—62) (8—33) .02 2 28.5 18.5 (14-162) (6-53) N.S. 3 22.0 14.5 (7.5—68) (8—50.5) N.S. 4 20.0 17.0 (6—45.5) (6—27.5) N.S. 5 33.0 34.5 (12-79) (5—47.5) N.S. 6 11.3 20.3 (6-66) (7.5-40) N.S 7 20.0 27.5 (10—116) (7.5-80) N.S. 8 14.5 10.0 (6.5-39) (5.5—22.5) N.S 9 19.5 14.5 (9—67.5) (4.5-32) N.S 102 which the lever correct on the prior trial was initially pressed by the mice on the subsequent trial. If the mice were noticing the relationship between some extraneous cue and the location of the correct box, I would expect them to return to this box on the next trial significantly more often than expected by chance. However, the actual and theoretical frequencies were not significantly different on seven of the nine cue sets. Curiously, statistical signifi— cance was obtained for the two cue sets having luminous circles. The mice returned to the box correct on the previ— ous trial significantly mpg; often than expected by chance Afor the object cue set 3 (t = +5.06, df = 5, p < .005), and significantly l§§§ often than expected by chance for cue set 7 (t = —4.69, df = 5, p < .02). The inconsistency in signs of the t values suggests that one or both of these results may have been obtained by chance alone. The mice's choices of the experimental visual cues can not be attributed to being able to see one but not the other of the two cues in a cue set. Table 12 suggests that the mice could clearly see all of the objects presented to them, for their errors on control trials were usually significantly greater than their errors on the other test situations. This was almost always true when controls were compared with opposition trials, and was sometimes true for the control— removal trial comparisons. Finally, the mice's choice behavior represents a delib— erate selection of the cues, because the mice were well 103 Table 12. One tailed probability"values for comparison of median errors in control versus removal and opposition trials. Numbers in parentheses refer to the specific cue set that these values are associated with. .Location numbers are those shown in Figure 2. Control Control vs. Removal Opposition Wilcoxon Mann- Wilcoxon Mann- Cue or Whitney or Whitney Walsh tests Walsh tests tests tests Horizontal tube (1) N.S. .05 .03 .008 Horizontal tube (9) N.S. .05 .03 .001 Gray Bowl (2) N.S. N.S. .05 .02 Gray Bowl (8) N.S. N.S. .025 .01 Luminous Circle (3) .025 .004 .025 .001 Luminous Circle (7) N.S. ‘N.s. N.S. .05 Vertical tube — Loc. 1 (4) N.S. N.S. .025 .02 Vertical tube Loc. 1 (8) N.S. N.S. .025 .002 Vertical tube Loc. 10 (3) .025 .01 .025 .002 Vertical tube Loc. 10 (5) N.S. N.S. .025 .001 Vertical tube Loc. 10 (6) N.S. N.S. .025 .001 Vertical tube Loc. 7 (l) .05 N.S. .03 .008 Vertical tube LOC. 7 (2) N.S. N.S. .03 .008 Vertical tube Loc. 7 (5) .025 .004 .025 .001 Vertical tube Loc. 3 (4) N.S. N.S. .03 .03 Vertical tube LOC. 3 (6) N.S. N.S. N.S. .002 Vertical tube LOC. 3 (7) N.S. N.S. N.S. .05 Vertical tube .3 Loc. 3 (9) N.S. N.S. .03 .008 .E 104 trained prior to testing. This is shown by their high per- formance level during testing (Table 13). This table con- servatively reflects performance. Not only are routine and opposition trials included, but also removal trials, which often had more errors than the other trials. Trials with only one object present (i.e., removal trials) account for almost 11% of the trials represented in this table. All statistical analySes for frequencies of various numbers of errors in Table 13 were performed with Wilcoxon tests, which paired the values for the nine cue sets. Combined frequencies for 0 and 1 errors were significantly greater than the combined frequencies of 2, 3, 4, and 2_5 errors (T = 0, p < .01). Since the two test days were not significantly different for either 0 or 1 errors in each of the cue sets, the two days were combined for this table. In terms of frequency, 0 errors > 1 error > 2 errors > 3 errors (for each comparison, T = O, p < .01)° However, there were no significant differences between the frequency of 3 and 4 errors, or 4 and 2,5 errors. Table 13. 105 on Which a particular number of errors was obtained. The sum of each row equals 100%. these values are associated with. Percentages of total number of trials (N = 228) The values are for test trials 1—19. The numbers in parentheses refer to the specific cue set that Number of errors per _ trial 0 l 2 3 4 25 HDrizontal tube vs. vertical tube (1) 84.6 8.3 4.8 0.8 1.3 0 Gray bowl vs. vertical tube (2) 59.2 22.4 7.9 4.4 3.1 3.1 Luminous circle vs. vertical tube (3) 65.3 19.3 8.3 2.6 1.8 2.6 Radial vs. peripheral displacement (4) 43.8 25.0 14.9 4.8 4.4 7.0 o 0 above box vs. 45° above box (5) 80.7 12.3 6.1 0.4 0.4 0 Peripheral vs. vertical displacement (6) 71.9 17.1 4.4 2.2 3.1 1.3 Luminous circle vs. vertical tube (7) 67.1 20.6 7.9 3.1 0.4 0.9 Gray bowl vs. vertical tube (8) 53.5 20.6 11.8 6.1 3.5 4.4 Horizontal tube vs. vertical tube (9) 69.7 19.7 3.9 3.5 0.9 2.2 DISCUSSION The visual cues g,.leucopus-prefers for orienting itself in its home range were studied in the laboratory by opposing cues which were abstractly analogous to some of those found in its natural woodland habitat. In this way, the laboratory results might be related to the.natural situation. Several general hypotheses and their fate on testing were as follows: 1) P. leuc0pus would prefer vertical ob- jects. The data supported this hypothesis on five of six cue sets. 2) P, leucopus would prefer objects as close as pos- sible to the goal. This hypothesis was not supported in one sense, for a cue zero cm from the correct box was not pre- ferred over a cue similarly placed but 38 cm above the box. However, this hypothesis was supported in the sense that objects having certain spatial co—ordinates in common with the goal were preferred. 3) g, leucopus} spatial preferences can be arranged in a logically consistent hierarchy. This hypothesis was supported, for the mice were most disoriented by a cue located radially near the arena's center, less dis- oriented by a cue peripherally placed and 450 from the goal, and still less disoriented by cues located peripherally and 00 from the goal. 4) When object preference is experimental- ly opposed to space location preference, one of these two 106 107 preferences would consistently take precedence over the other in all cue sets and by all mice. This hypothesis was partial- ly supported. Object preference took precedence on two of the three cue sets by most of the mice. The principle aim of this study was to evaluate the relative preference for various visual cues and their loca- tions, and then to relate these findings to mouse orienta- tion in the natural woodland habitat. The cues and their various stimulus aspects can be arranged in a preference ‘ hierarchy that reflects the priority with which they were used in the laboratory orientation task. Since the numbers of objects and spatial positions were limited in this study, the hierarchy is not complete and the position of various factors in the hierarchy could change when more information is available. This hierarchy is based on certain key rela— tionships between cue sets rather than on some frequency measure. In preview, the hierarchical arrangement is as follows: 1) parallax; 2) verticality; 3) horizontality; 4) contrast with background; 5) spatial contiguity, shape, and size (all seem to have about the same rank in the hier— archy); 6) luminosity; 7) shadows. The hierarchical arrange- ment and evidence for and against the specific location of a factor in the hierarchy is considered in detail bélOw. l) Parallax: A mouse usually oriented to a cue that was unambiguously associated with the goal box regardless of his position in 108 the arena. If a cue's relationship to the goal changed with his movement about the arena, then he usually oriented to the other cue which did not change in relationship to the goal (that is, did not exhibit parallax). For example, mice preferred the vertical tube over the horizontal tube, except when the vertical tube was 450 counterclockwise from the goal; then, the horizontal tube directly on top of the box was preferred. Although it could be argued that the hori- zontal tube was selected because it was nearest to the goal (spatial contiguity), relative distances to the goal were not critical, for mice exhibited pg preferences for a cue zero cm from the goal (on the box) over the cue 38 cm above the goal. Also, when identical objects were located 38 cm equidistant from the goal, both above and beside it, no preference should be shown if spatial contiguity was such a critical factor. Instead, the mice preferred the cue loca— tion which did not exhibit parallax (above the box) over the location which did exhibit parallax (450 from the box). Further evidence for my placing parallax on the top of this hierarchy is that the median number of days to reach final training stage criterion was greater when both locations involved parallax (set 4, 6 days), than when at least one location did not involve parallax (other Phase II sets, 2-2.5 days). 2. leucopus may have chosen a vertical tube 450 from the correct box over a vertical tube near the arena center 109 (set 4) because parallactic displacement may have been less extreme for the 450 location than for the near center loca- tion as a mouse moved around the arena. However, this paral- lactic displacement factor is not the only explanation for the choice. An alternative explanation is that a mouse concentrates its attention on the arena's periphery, the location of the goal box. Neither the relative extent of parallactic displacement or attending to the periphery can explain the preference for vertical tube near the arena center over gray bowl at the periphery (set 8). Here, verticality exerted more control over the orientation of the mice than did either parallax or attending to the arena periphery. There'was one exception to the general principle of preference for situations not involving parallax. Mice pre— ferred a vertical tube 450 from the goal, exhibiting parallax, over a luminous circle directly above the goal, not exhibit— ing parallax. However, as noted in the preview of the hier— archy, luminosity was assigned to the bottom of the hierarchy (a luminous object was never preferred when paired against another object with the same spatial co—ordinates). Nevertheless, one of the mice (number 6) chose in accordance with the parallax principle, for it preferred the luminous circle above the goal. Parallax occurs at the top of this hierarchy because it was the principle least violated by the orienting mice. They required a cue whose spatial association with the goal did not change as they moved about the arena. 110 2) Verticality: The mice oriented to a vertical tube in preference to the other three objects, and even oriented to a vertical tube occupying an unpreferred spatial location (with the ex- ception of the one cue set discussed in the parallax section). P, leucopus may have so consistently preferred a vertical object because it had the same shape from any view— point in the arena (shape invariance), and was conspicuous (that is, had a small basal area relative to height). Although vertical objects were present in all cue sets, prior exposure to them can not explain the obtained choice results for reasons that will be considered in detail later when .prior exposure effects are discussed. Verticality (and horizontality) have also been implicated as primary factors in visual perception for another nocturnal rodent, the rat (Dodwell and Niemi, 1967). 3) Horizontality: Although not directly compared, a horizontal object was probably preferred over all other objects except the vertical object, for g. leucopus showed a smaller difference in preference for vertical over horizontal than for the other objects (see Figures 5 and 6; Table 4). In fact, the horizontal object was actually preferred when it did not exhibit parallax and the vertical object did exhibit parallax. 111 The mice may have preferred a horizontal object some— what less than a vertical object because the horizontal object varied in its apparent shape depending on a mouse's location in the arena, and it was less conspicuous. The choice was probably not made on the basis of background contrast, for both black cues had their total perimeters contrasting with the background gray arena wall. Nor was the choice made on the basis of slight differences in the cue locations atop the box; their exact geometrical loca— tions did not significantly influence choices, as revealed by the Friedman test. Horizontality appears near the top of this spatial hierarchy because of the relationship of the horizontal object to the high-ranking vertical object. 4) Contrast with background: Similarly, contrast with background was assigned its location in the hierarchy because of the relationship of a gray bowl, which had low contrast (other attributes will be considered later) to a vertical object which had high con- trast. The mice usually chose the vertical object on more trials when it was opposed to the gray bowl than when it was Opposed to the horizontal tube (Table 4). Also, vertical object choices with the gray bowl present were significant at lower probability values. The relationship between hori— zontality and low contrast is also suggested by the number of days to reach the final training stage criterion: 5.5 days for the cue set with a horizontal (high-contrast) object, 112 but 7 days for the low-contrast object. This relationship is also implicated by the Phase III results. The mice chose a vertical object over a gray bowl in.a.preferred spatial location but when the horizontal tube was in a preferred location, they chose it rather than the vertical Object. The relatively low hierarchical position for contrast is also suggested by a pilot study result. Mice did not prefer vertical stripes over a rectangle, even though the vertical stripes had a greater total length of black—gray contrast (Table 2, set 4a). Nevertheless, assignment of contrast to fourth place in this hierarchy involved a high degree of uncertainty. It could be argued that contrast should occupy a higher place in this hierarchy since the highly preferred vertical object itself had high contrast with background. But, as previously discussed, contrast differences can not explain a vertical object being chosen over a horizontal object. The relative role of contrast per sé could be resolved by future experi- mentation, such as pairing a vertical object (medium contrast) with a horizontal object (high contrast). 5) Spatial contiguity, shape, and size: These factors all occupy about the same low hierarchical position. Alone, they can not readily explain any of the definitive series results. Their hierarchical position is somewhat indeterminate, but there is no evidence to suggest they are avoided (as may be true for luminosity). 113 Distance from the goal box (spatial contiguity) does not adequately explain choice behavior, for the mice showed no preference for a cue zero cm from the goal (on the box) over a cue 38 cm above it; this implies that the mice have a.large visual field within which they will use cues for orienting. Shape preferences do not adequately explain the re— sults, although shape invariance seemed to be important. Mice preferred vertical over horizontal tubes although both were identical in shape. Their preference for a vertical tube over gray bowl and luminous circle was probably pg; based on the circular shapes of these two objects. Mice in the third pilot study showed no preference between a rec- tangle and two circles (Table 2, set 4b), and preferred a luminous circle over approximately rectangular cues ("canopy", set 9). g. leucopus preferred black or white circles over a luminous circle of the same size and space location (sets 6 and 7). This suggests that the vertical tube preference over luminous circle may have been based on nonluminosity (and verticality) rather than rectangularity versus circu— larity. Size differences also do not adequately explain the results. The horizontal and vertical tubes were both the same size, yet vertical tube was definitely preferred. Vertical tube was also preferred over larger objects (gray bowl, and luminous circle--the latter being larger in 114 diameter). In the space preference phase, the preferred cue was of the same apparent size as viewed from the arena's center (set 6) or perhaps even smaller (set 4). 2,.leucopus did not prefer a cue on the box over a cue above the box (set 5) although there may have been an apparent size dif- ference between the two cues. Size, shape, and spatial contiguity factors might in- fluence P, leucopus orientation considerably more than was evident for the particular cue sets used in this study. Further experimentation could more accurately define their roles. For example, size might exert considerable control on mouse orienting preference if a thick and long horizontal tube were paired with a thin and short vertical tube. 6) Luminosity: Luminosity is ranked lower in the hierarchy than the above mentioned factors because the mice appeared to avoid the luminous cue. Although hardly conclusive, the following evidence suggests this possibility: l) g, leucopus oriented to extraneous cues more often than would be expected by chance when its orientation task included a luminous circle in Phase I. 2) The mice had removal trial latencies for the luminous circle which were significantly longer than those for the vertical tube in Phase III (Table 9). 3) P, leucopus needed 8.5 days to reach the last training stage criterion for the luminous circle cue set in Phase I as compared with values ranging between 2 and 7 days for the other eight cue sets. 115 In any case, luminosity is certainly not preferred, if not actually avoided. The mice usually chose the vertical object on more trials when it was opposed to the luminous circle than when it was opposed to the other Phase I objects (Table 4). They also chose the vertical tube over the luminous circle (Table 8) even when the vertical tube exhi— bited parallax and the luminous circle did not exhibit parallax (set 7). The lack of preference by the mice for the luminous circle was probably based on this object's luminosity attri— butes. It had considerable contrast with the background, so the mice probably did not choose on the basis of contrast. Nor were their preferences based on shape or size differs ences, for reasons previously stated It is unlikely that their dark-adapted vision was im- paired for the following reasonszi l) Vestal (1970) found no significant effect of bright light illumination on P. leucopus' visual acuity. 2) The luminous circle over the correct box increased floor light levels less than three percent. 3) Their luminous circle opposition and removal trial errors and latencies were usually not significantly different from those for the vertical tube (Tables 5 and 9). 4) Their dim-light control trial errors for luminous circle cue sets were significantly larger than their opposition or removal errors (Table 12). 5) The frequency distribution of test trial errors for luminous circle cue sets were com- parable to those for other cue sets (Table 13). 116 7) Shadows: Although the cues cast shadows, it is unlikely that P..leuc0pus oriented to shadows. In the third pilot study, P. leucopus preferred the low-ranking luminous circle over a strong shadow-casting canOpy cue (Table 2, set 9), and preferred a vertical tube over a horizontal tube which cast a strong shadow leading directly from tunnel exit to goal box (set 5). Shadows cast by theiexperimental series cues were not nearly so pronounced as were these shadows. Table 14 summarizes the major evidence used for assign— ing each factor to a particular position in the hierarchy. Each cue set, and the factors which influenced choice, is then considered. The following ligp summarizes for each cue set the major factors which probably influenced the cue choice by the mice. 1. Horizontal tube versus vertical tube (set 1). Verticality (perhaps related to shape invariance and conspicuousness). 2. Gray bowl versus vertical tube (set 2). Verticality and contrast with background. 3. Luminous circle versus vertical tube (set 3). Verticality and nonluminosity. 4. Near the arena's center versus 450 from the box (set 4). Extent of parallax, and attention focused on arena's periphery. 117 5. On the box versus over the box (set 5). Large visual field (pg; spatial contiguity). 6. Over the box versus 450 from the box (set 6). Lack of parallax. 7. Vertical tube 450 from the box versus luminous circle over the box (set 7). Verticality and nonluminosity override parallax. 8. Vertical tube near the arena's center versus gray bowl 450 from the box (set 8). Verticality and contrast with background over- ride extent of parallax and attention focused on arena's periphery. 9. Horizontal tube on the box versus vertical tube 45° from the box (set 9). Lack of parallax overrides verticality. What mechanisms could account for the above preference hierarchy of factors which the mice used in orienting? Three mechanisms could be proposed: 1) exposure to the same cue on an earlier cue set (prior exposure), 2) early experience, and 3) genetic control. Prior exposure to a cue could bias preference on a subsequent cue set. Certainly such bias could occur in the experimental series particularly for vertical objects, which were present in all cue sets. Some of the aberrant choices by individual mice may be attributable to prior exposure to a cue. 118 Table 14. Summary of evidence for a preference hierarchy. Factor Key Evidence 1. Lack of parallax 1. Horizontal tube (0°)preferred over vertical tube (45°). 2. Longest number of training days when both cues exhibited paral- lax. 2. Verticality Vertical objects preferred on five of six cue sets (see 1.). 3. Horizontality Vertical object preferred over horizontal object (p = .05+, chi square). 4. Contrast (hierarch— Vertical object preferred over a ical position un— certain) low contrast object (p <3.001, chi square). Spatial contiguity, shape, and size (hierarchical posi— tion uncertain). 1. Alone, they can not readily explain any of the definitive series results. However, they were not avoided. Luminosity Some weak evidence for avoid- ance. The mice usually chose the verti— cal object on more trials when it was opposed to the luminous circle than when it was opposed to the other Phase I objects. They preferred the vertical object even when it exhibited parallax and the luminous circle did not exhibit parallax (Phase III). 7. Shadows Preliminary series: Luminous circle preferred over a shadow- casting canopy cue. 119 However, the following evidence suggests that bias from prior exposure was minimal in the experimental series: 1) The mice showed no significant differences in magnitude of choice for a cue regardless of whether the cue appeared in the first, second, or third cue set in a phase (Friedman test). If mice were choosing on the basis of prior exposure to‘a cue, I would expect the frequency of their choices to be least for the first set and greatest for the third set. 2) The mice showed consistent differences in the frequency of their choices for various cue sets. For example, most mice chose the vertical object on more trials when it was opposed to the luminous circle than when it was opposed to the gray bowl (Table 4). 3) All mice preferred a horizontal tube on tOp of the box over a vertical tube 450 from the box, even though they had been exposed to vertical objects on all prior cue sets. 4) Mice usually showed the same cue choices in both the third pilot study and the experimental study even though the two differed quite markedly in the types and temporal ordering of their cue sets. For example. P. leuCOpus preferred vertical tubes over horizontal tubes in both studies. If prior exposure to test cues does not account for the cue choices, what mechanisms could account for them? Early experience with very similar objects seems unlikely, for the mice were laboratory-reared. However, stimulus general- ization from early exposure to horizontal and vertical 120 surfaces, to contrast differences, to lights, etc., may account for some results. An alternative explanation is that the preferences were genetically controlled. Such perceptual preferences have been demonstrated for Peromyscus habitat selection (Harris, 1952; Wecker, 1963). These two alternatives could be experimentally investigated by species hybridization studies and systematic variation of early experiences. In either case, the mice may have selected the cues on the basis of their being representatives of general classes or on the basis of the specific attributes of the indi- vidual cues. For example, the vertical tube may have been chosen over the gray bowl not on the basis of preference for vertical cues per se but simply because of the difference in contrast between the two objects and their backgrounds. These two alternatives could be experimentally resolved by systematically changing various attributes of the cues which were opposed for choice, and by using a number of different representatives of a given cue class. The results have so far only been discussed in terms of one measure. Several other measures were taken in addition to the opposition choice frequency measure, which provided the most direct method of determining preferences. The re— sults for the other measures confirmed the value of the choice frequency measure because it was the only one that provided statistically significant differences. The other 121 measures failed to indicate the direction of choice. The median number of errors made on removal trials (only one cue present) confirmed the opposition choice frequency in only four of nine cue sets. Similarly, the removal trial latencies to obtain water confirmed the frequency measure in only five of nine sets. Opposition trial errors confirmed the frequency measure on only two of nine sets. Two explanations can be offered for the failure of error and latency scores to indicate cue choice. The simplest explanation is that these dependent variables were too easily influenced by extraneous factors such as intermittent en- vironmental noises, changes in the subjects' motivation, and temperamental differences between subjects. If such factors were influential, I would expect the absence of any system- atic differences from one cue set to the next in the error and latency measures. A comparison of the median values for the most and least preferred cue in each phase revealed no significant differences for any of the measures, which indi- cates that extraneous factors could have randomly contributed to these measures. The second explanation is that the latency and error scores measured cue discriminability ("noticeability"), and the cues may have all been about equally discriminable. Even though two cues are equally discriminable, a subject may still prefer one of them (Trabasso and Bower, 1968), as indicated by the choice frequencies on opposition trials in 122 this study. That the mice could see and did notice at least one of the two cues is evident from the significantly greater number of errors made during the control trials with the light intensity reduced. Within each cue set, the cues may have been about equal in discriminability, for 1) both cues had overall medians of zero errors on six of nine sets, and 2) removal trial errors for all cues were not signifi— cantly different from the routine fourth trial's errors (both cues present). Thus, the failure of error and latency scores to indicate cue choice may be-attributable to either one or both of two explanations: extraneous factors were influential, and the cues were about equally discriminable. What could the cue choices and the hierarchy established here imply for the orientation of g. leucopus in its natural woodland home range? The results suggest that P. leucopus orientsprimarily to trees proximally associated With its goal. In many instances the goal may be-thetree itself, for trees provide nest sites (Nicholson, 1941), food (NicholSon, 1941; Hamilton, 1941), andpotential escape from predators (Smith and Speller, 1970). Although not yet experi— mentally demonstrated, it is likely that g. leucopus also selects its habitat using prominent vertical cues; this has been demonstrated for P, maniculatus gracilis (Harris, 1952), a lgsg arboreal species than P. leucopus (Smith and Speller, 1970). Field observations support the speculation, for this species usually ran to the base of a nearby tree on live 123 trap release (Smith and Speller, 1970). Personal observa— tions based on snow—tracking also suggest orientation ~primarily'to‘trees. Seventy percent of P, leuCOpus track sets (N = 39) ended at a large tree. The track sets were quite linear, which further suggests direct orientation to the large tree at the end of the tracks. On seven track sets, 3, leucopus ran directly to a tree with no other object within two feet of each side of the tracks. 'Similarly, in the laboratory arena, 3. leuCOpus usually ran in a straight line to the goal box and its associated cues. The lack of preference for a near-floor level cue over a cue elevated 450 above the correct.box implies that g. leucppus orients-not only to a tree's base but also to a considerable portion of its trunk, which has the preferred property of verticality as well. The peripheral displacement cue sets suggest that g. leucopus can be guided to a goal by a tree located some dis— tance from it, if no prominent object is proximally associ— ated with its goal. Objects located between the mouse and its goal are probably little used in orienting because the mouse would soon leave the cue behind as it progressed guideless toward the goal. The radial displacement results, which implicate these tactics, may have been biased, however, by the short distance (about a mouse's body length) between the cue and the onset of the mouse's arena travels. Orientation to 124 objects located between the beginning and apparent goal of tracks inthe snow indicated that a mouse may obtain its bearings from cues near the onset of its journey. It could use these cues for orienting by two different tactics. The simplest tactic would be for the mouse to orient visu— ally to cues along its path and, after reaching each cue, reorient visually to the next cues in line until the goal is reached. Conversely, it might orient visually to the cues close by and then maintain the same path direction to the goal by some inertial process. These alternatives could be experimentally tested by removal of the objects associ- ated with the goal. If an inertial process is involved, the mouse would still run directly to the goal. Secondary cue preferences of P. leucopus could be fallen logs (horizontal objects). If a fallen log is considerably closer to the goal than a tree, the log would be used for orientation. Similarly, fallen limbs of trees, brush piles, horizontal fissures in rock strata, and perhaps even a shore line could provide suitable orientational cues. Mounds of earth, or rocks, similar to this study's gray bowl in their shape and background contrast, are probably infrequently used as orientational cues. Distant hills and other objects similar to the gray bowl in shape, but having somewhat more contrast with background, are probably less used in orientation than distant trees on the horizon. 125 The moon, a luminous circle, is probably-not used as an orientation cue. Kavanau (1968) has reached the opposite conclusion for Peromyscus on the basis of running wheel activity with a small directional light source present in totally dark surroundings. However, slightly less than half his subjects oriented to the light source, some later re- versed their light-orienting direction, and the light source was not paired with any other discrete object to assess cue preference. Consistent orientation to the moon would be difficult, for the moon's appearance and location continually change. Clouds and vegetation would frequently obscure the moon for this species of mouse. Also, light from a full moon inhibits g, maniculatus activity (Falls, 1968). A clear preference for some visual cues over others should facilitate terrain "familiarization" within or out- side the home range. Indirect and zigzagging routes of homing mice may represent orientation to conspicuous pre— ferred objects previously encountered in familiar terrain. Although Peromyscus orientation has been extensively studied by field release methods, the cues used in homing are still unknown. This study has departed from prior studies by demonstrating, in a laboratory situation, the visual cue preferences for orientation by one species of Peromyscus. Since these cues may be analogous to natural objects, this study may also suggest the visual cues used by 126 g..leucopusin orienting within its home range. At present, however, this is only an assumption. It needs verification by controlled manipulative field Studies, which may require some training of the mice and the use of objects as similar as possible to trees and 109s. Such studies are to be pre- ferred over naturalistic field studies whose data could be subject to diverse interpretation. Successful replication with a field manipulative study would give some assurance that subsequent laboratory studies were indeed related to home range orientation. These studies could include investi- gations of the specific attributes of the top-ranked cues in the hierarchy, broadening the hierarchy by studying more cues, or assessing the relative roles that genetics and experience play in determining the hierarchy. SUMMARY 1. g. leucopus males were tested for visual cue prefer— ences used in orientation by opposing two cues which the mice had been trained to use in reaching a goal. 2. The mice oriented to visual aspects of the cues rather than to their nonvisual aspects, could see each cue, and did not orient to extraneous cues. 3. g. leucopus preferred a vertical tube over all other objects. 4. Spatial displacement preferences can be arranged in a descending hierarchy, with the most preferred location directly over the goal. In terms of cue displacement, radial displacement affected mouse orientation most (peripheral location preferred), peripheral displacement affected orien— tation less (00 preferred), and vertical displacement had no effect on orientation (neither cue preferred). 5. A hierarchy of factors which may have contributed to cue selection, arranged in descending order, was pro— posed: parallax; verticality; horizontality; contrast with background; spatial contiguity, shape, and size; luminosity; and shadows. 127 1:} 128 6. In its woodland home range, g. leucopus probably orients primarily to trees in proximate association to its goal, orients secondarily to fallen logs, rarely orients to hills, mounds of earth or rocks, and never uses the moon for orientation. hall; BIBLIOGRAPHY BIBLIOGRAPHY Beer, J. R. 1961. Winter home ranges of the red-backed mouse and white-footed mouse. J. Mamm. 42: 174—180. Berry, R. M., Rogers, R. L. and J. E. McCarrol, 1968. Spontaneous pattern discrimination in the rat: a further analysis. Psychon. Sci. 12: 100. Blair, W. F. 1940. A study of priarie deer-mouse popula- tions in southern Michigan. Amer. Midl. Nat. 24: 273—305. Blair, W. F. 1951. Population structure, social behavior and environmental relations in a natural population of the beach mouse (Peromyscus polionotus leucocephalus). Contr. Lab. Vert. Biol. Univ. Mich. 48: 1-47. Bliss, C. I. 1967. Statistics in biology.' Vol. 1. McGraw-Hill, New York. Bovet, J. 1960. Experimentslle Untersuchungen fiber das HimfindevermOgen von Mausen. Z. Tierpsychol. 17: 728-755. , Bovet, J. 1962. Influence d'un effet directionnel sur le retour au gite des mulots fauve et sylvestre (Apodemus flavicollis Melch et A. sylvaticus-L.) et du campagnol roux,(Clethrionomys glareolus Schur.) (Mammalia, Rodentia). Z. Tierpsychol. 19: 472—488. Bovet, J. 1965a. Note sur le retour au gite du Mulot en Camargue. Z. Tierpsychol. 22: 1963-166. Bovet, J. .1955b. Ein Versuch, wilde mause unter ausschluss optischer, akustischer und osmisher Merkmale auf Himmelsrichtungen zu dressieren. Z. Tierpsychol. 22: 839—859. » Bovet, J. 1968. Trails of deermice (Peromyscus maniculatus) traveling on the snow while homing. J. Mamm. 49: 713-725. Bovet, J. 1970. Sedentariness of Peromyscus maniculatus in Alberta granaries. J. Mamm. 51: 632—634. 129 130 Bovet, J. 1971. Initial orientation of deer mice (Peromys— cus maniculatus) released on snow in homing experiments. Z. Tierpsychol. 28: 211—216. Brotzler, A. 1963. Untersuchungen zum Orientierungsver— halten des Syrischen Goldhamsters (Mesocricetus auratus Waterhouse). Z. Vergl. Physiol. 47: 148-190. Burt, W. H. 1940. Territorial behavior and populations of some small mammals in southern Michigan. Mis. Publ. Mus. of Zool. Univ. Mich. No. 45. Calhoun, J. B. 1962. The ecology and sociology of the Norway rat. U.S.D.A.H.E.W., Publ. Health Service, viii + 288 pp. Chitty, D. 1937. A ringing technique for small mammals. J. Anim. Ecol. 6: 36—53. Cox, D. R. 1958. Planning of experiments. John Wiley and Sons, Inc., New York. Davis, R. 1966. Homing performance and homing ability in bats. Ecol. Monogr. 36: 201~237. Dice, L. R. and E. Barto. 1952. Ability of mice of the genus Peromyscus to hear ultrasonic sounds. Science 116: 110—111. Dice, L. R., Barto, E. and P. J. Clark. 1963. Modifications of behavior associated with inherited convulsions or whirling in three strains of Peromyscus. Anim. Behav. 11: 40—50. Dodwell, P. C. and R. R. Niemi. 1967. Contour separation and orientation in shape discrimination by rats. Psychon. Sci. 9: 519—520. Falls, J. B. 1968. Activity. In: J. A. King (ed.), Biology of Peromyscus (Rodentia). Spec. Publ. No. 2, Amer. Soc. of Mamm. Fisler, G. F. 1962. Homing in the California vole, Microtus californicus. Amer. Midl. Nat. 68: 357—368. Fisler, G. F. 1966. Homing in the western harvest mouse, Reithrodontomys megalotis. J. Mamm. 47: 53—58. Fisler, G. F. 1967. An experimental analysis of orienta— tion to the homesite in two rodent species. Canadian J. Zool. 45: 261—269. 131 Gentry, J. B. 1964. Homing in the old—field mouse. J. Mamm. 45: 276-283. Gibson, E. J. 1969. Principles of perceptual learning and development. Appleton-Century Crofts, New York. Gould, E., Negus, N. C. and A. Novak. 1964. Evidence for echolocation in shrews. J. Exp. Zool. 156: l9—36. Gould, E. 1965. Evidence for echolocation in the tenrecidae of Magagascar. Proc. Amer. PhilOSOph. Soc. 109: 352— 360. Griffin, D. R. and R. J. Hock. 1949. Airplane observations of homing birds. Ecology 30: 176—198. Griffo, J. V., Jr. 1961. A study of homing in the cotton mouse, Peromyscus gossypinus. Amer. Midl. Nat. 65: 257-289. Grunwald, A. 1969. Untersuchungen zur Orientierung der Weisszahnspitzmaeuse (Soricidae: Crocidurinae). Z. vergl. Physiol. 65: 191—217. Hamilton, W. J., Jr. 1937. Activity and home range of the field mouse, Microtus p. penngylvanicus (Ord.). Ecology 18: 255-263. Hamilton, W. J., Jr. 1941. The food of small forest mammals in eastern United States. J. Mamm. 22: 250—263° Harris, V. T. 1952. An experimental study of habitat selec— tion by prairie and forest races of the deermouse, Peromyscus maniculatus. Contr. Lab. Vert, Biol. Univ. Mich. 56: 1w53. Horner, B. E. 1954. Arboreal adaptations of Peromyscus, with special reference to use of the tail. Contr. Lab. Vert. Biol. Univ. Mich. 61: 1-84. Howard, W. E., Marsh, R. E. and R. E. Cole. 1968. Food detection by deermice using olfactory rather than visual cues. Anim. Beh. 16: 13-17. Karmel, B. Z. 1969. Complexity, amounts of contour, and visually dependent behavior in hooded rats, domestic chicks, and human infants. J. Comp. Physiol. Psychol. 69: 649-657. Kavanau, J. L. 1967. Behavior of captive white—footed mice. Science 155: 1623-1639. Kavanau, J. L. 1968. Activity and orientational responses of white-footed mice to light. Nature 218: 245—252. 132 King, J. A., 1968. Psychology. In: J. A. King (ed.), Biology of Peromyscus (Rodentia). Spec. Publ. No. 2, Amer. Soc. of Mamm. King, J. A. 1970. Ecological psychology: an approach to motivation. Nebr. Symp. on Motivation, 1—33. Law, 0. T. 1954. Preference in the rat for vertical or horizontal stripes after training on a white—black discrimination. Amer. J. Psychol. 67: 7l4~7l6. Lay, D. W. and R. H. Baker. 1938. Notes on the heme range and ecology of the Attwater wood rat. J. Mamm. 19: 418—423. Layne. J. N. 1954. The biology of the red squirrel, Tamiasciurus hudsonicus loguax (Bangs) in New York Ecol. Monogr. 24: 227~267. Layne, J. N. 1957. Homing behavior of chipmunks in central New York. J. Mamm. 38: 519—520. Lehmann, E. v. 1956. Heimfindeversuche mit kleinen Nagern. Z. Tierpsychol. 13: 485—491. Lindenlaub, E. 1955. Uber das HeimfindevermOgen von saugetieren. II: Versuche an Mausen. Z. Tierpsychol. 12: 452—458. Lindenlaub, E. 1960. Neue Befunde fiber die Anfangsorien— tierung von Mausen. Z. Tierpsychol. 17: 555—578. Lfiters, W. and G. Birukow. 1963. Sonnenkompassorientierung der Brandmaus (Apodemus agrarius Pall.). Naturwiss. 50: 737—738. McCabe, T. T. and B. D. Blanchard. 1950. Three species of Peromyscus. Rood Associates, Santa Barbara. Meier, G. W. 1964. Differences in maze performances as a function of age and strain of housemice. J. Comp. Physiol. Psychol. 58: 418—422. Metzgar, L. H. 1967. An experimental comparison of screech owl predation on resident and transient white—footed mice (Peromyscus leucopus). J. Mamm. 48: 387—391. Meyer, D., Treichler, F. and P. Meyer. 1965. Discrete- trial training techniques and stimulus variables. In: Schrier, Harlow, and Stollnitz (eds.), Behavior of nonhuman primates, Vol. I. Academic Press, New York. 133 Moody, P. A. 1929. Brightness vision in the deer-mouse, Peromyscus maniculatus gracilis. Exper. Zool. 52: 367-405. Morris, R. D. 1967. A note on the homing behavior of Peromyscus maniculatus osgoodi. Canadian Field Natur. 81: 225—226. Munn, N. L. 1950. Handbook of psychological research on the rat. Houghton Mifflin Co., Boston. Murie, O. J. and A. Murie. 1931. Travels of Peromyscus. J. Mamm. 12: 200-209. Murie, M. 1963. Homing and orientation of deermice. J. Mamm. 44: 338-349. Nicholson, A. J. 1941. The homes and social habits of the wood-mouse (Peromyscus leucopus noveboracensis) in southern Michigan. Amer. Midl. Nat. 25: 196—223. Oliverio, A. and D. Bovet. 1966. Effects of age on maze learning and avoidance conditioning of mice. Life Sciences 5: 1317524. Orr, H. D. 1959. Activity of white-footed mice in rela— tion to environment. J. Mamm. 40: 213-221. Polidora, V. J. 1965. Personal communication to J. A. King. Rawson, K. S. 1956. Homing activity and endogenous activ— ity rhythms. Ph. D. Dissertation. Harvard University. 111 pp. Rawson, K. S. 1966. Goal directed orientation in the homing behavior of mice (genus Peromyscus). Bull. Ecol. Soc. Amer. 47: 199. Rawson, K. S. and P. H. Hartline. 1964. Telemetry of homing behavior by the deermouse, Peromyscus. Science 146: 1596-1598. Riley, D. A. and M. R. Rosenzweig. 1957. Echolocation in rats. J. Comp. Physiol. Psychol. 50: 323—328. Robinson, W. L. and J. B. Falls. 1965. A study of homing of meadow mice. Amer. Midl. Nat. 73: 188-224. Schleidt, W. 1951. Orientierende Versuche uber die ‘Heimkehrfahigkeit der Rotelmaus (Evotomys glareolus ruttneri Wottst.). Z. Tierpsychol. 8: 132-137. 134 Schmidt-Koenig, K. 1965. Current problems in bird orien— tation. In: D. S. Lehrman, R. A. Hinde, and E. Shaw (eds.). Advances in the study of behavior. Academic Press, New York. Sheppe, W. 1965a. Dispersal by swimming in Peromyscus leucopus, J. Mamm. 46: 336-337. Sheppe, W. 1965b. Characteristics and use of Peromyscus tracking data. Ecology 46: 630—634. Sheppe, W. 1966. Determinants of home range in the deer mouse, Peromyscus leucopus. Proc. Calif. Acad. Sci. 34: 377-418. Smith, D. A. and s. W. Speller. 1970. The distribution and behavior of Peromyscus maniculatus gracilis and Peromyscus leucopus noveboracensis (Rodentia: Cricetidae) in a southeastern Ontario woodlot. Canadian J. Zool. 48: 1187~ll99. Stickel, L. F. 1949. An experiment on Peromyscus homing. Amer. Midl. Nat. 41: 659—664. Stickel, L. F., and O. Warbach. 1960. Small—mammal popu— lations of a Maryland woodlot, 1949—1954. Ecology 41: 269—286. Stickel, L. F. 1968. Home range and travels. In: J. A. King (ed.), Biology of Peromyscus (Rodentia). Spec. Publ. No. 2, Amer. Soc. of Mamm. Sutherland, N. S. 1961. Shape discrimination by animals. P S. Monogr. No. l. Terman, C. R. 1962. Spatial and homing consequences of the introduction of aliens into semi—natural populations of prairie deer—mice. Ecology 43: 216-223. Thomsen, H. P. 1945. The winter habits of the northern white—footed mouse. J. Mamm. 26: 138-142. Trabasso, T. and G. H. Bower, 1968. Attention in learning: theory and research. Wiley and Sons, Inc., New York. Vestal, B. M. 1970. Development of visual acuity in two species of Peromyscus (Rodentia). Ph. D. Dissertation, Michigan State University. 72 pp. Wecker, S. C. 1963. The role of early experience in habitat selection by the prairie deer mouse, Peromyscus maniculatus bairdii. Ecol. Monogr. 33: 307-325.