Journal of Experimental Psyhology: Animal Behavior Proesses 1999, Vol. 25, No. 2,153-167 Copyright 1999 by the Amerian Psyhologial Assoiation, In. 0097-7403/99/$3.00 Inentive Downshifts Evoke Searh Repertoires in Rats Norman C. Peoraro, William D. Timberlake, and Matthew Tinsley Indiana University Bloomington _ Negative inentive ontrast effets (NCEs) have typially been attributed to frustration or the deremental generalization of learned assoiations. The purpose of these experiments was to larify the relation of NCEs to the repertoires of funtional searh behaviors evoked by inentive downshifts. Rats shifted from 32% to 4% surose-solution dereased onsummatory responses but inreased nose-down loomotion, orientation, loation entries, and sampling of alternatives relative to unshifted ontrols. These hanges in behavior were terminated or failed to our under inentive upshifts. Furthermore, reward downshifts did not produe avoidane of the loation of the shifted inentive. Inreased searh ourred whether or not alternative reward loations were available. Together the evidene suggests that NCEs are related to evoked searh modes supporting a repertoire of funtional behaviors related to finding food. Negative ontrast effets (NCEs) are abrupt hanges in behavior that usually follow a disriminable redution in the austomed size or quality of reward. These hanges have been of partiular interest beause they are abrupt, exaggerated, and have been assumed to have an emotional basis. The abruptness of the behavioral hanges suggests little dependeny on habit formation. The behavioral hanges are onsidered exaggerated beause they are larger than would be expeted if the animals simply adjusted to the new level of reward (i.e., adopted the behavior of unshifted animals never experiening the larger reward). For example, ompared with unshifted rats always reeiving small rewards, rats shifted from large to small rewards show lower intake of foods (Flaherty, 1982) and slower approah speeds in runways (Crespi, 1942). Changes in behavior also appear to be emotion based in that rats may show esape or withdrawal from the reward loation (Daly, 1974a, 1974b) and avoidane in the form of inreased hoie errors on a maze (Elliott, 1928). The dominant explanation for NCE has foused on emotional onstruts suh as disappointment or frustration (Amsel, 1951; Crespi, 1942; Spene, 1956; Tinkelpaugh, 1928; see Flaherty, 1982 and 1996 for omprehensive reviews). Many experiments support the frustration explanation of reward-shift effets in runway situations (Amsel, Norman C. Peoraro, William D. Timberlake, and Matthew Tinsley, Department of Psyhology and Center for the Integrative Study of Animal Behavior, Indiana University Bloomington. This work was supported by National Institute of Mental Health Grant 37892 and National Siene Foundation (NSF) Grant IBN-9408366m and by Researh Assistantship and Researh Support Awards from the Program in Animal Behavior at Indiana University (NSF Grant BIR-9413220). We thank Rory Gont, Christine Clutter, Christopher Kurtz, and Mihael Raza for their assistane in these experiments. The first author thanks Beatrix T. and R. Allen Gardner for their sound advie. Correspondene onerning this artile should be addressed to Norman C. Peoraro, Department of Psyhology and Center for the Integrative Study of Animal Behavior, Indiana University, Bloomington, Indiana 47405. Eletroni mail may be sent to npeorar@indiana.edu. 1992). Behaviorally, failure to find expeted rewards or signals prediting the absene of reward has produed withdrawal from the loation assoiated with the redued reward (Adelman & Maatsh, 1956; Hearst, Bottjer, & Walker, 1980) and has energized the rate of subsequent instrumental behavior (Amsel, 1962). Physiologially, ues assoiated with dereases in reward have been shown to inrease irulating ortiosterone, a omponent of the adrenal stress response (Coover, 1983; Flaherty, Beker, & Pohoreky, 1985), whereas administration of anxiolyti drugs an derease NCE (Flaherty, 1991). However, there are reasons to question the ompleteness of a frustration aount. At a behavioral level, the form of frustration responses is not well speified in advane. The definition of esape responses as leaving the loation of the downshifted reward is operationally lear but open to the alternative interpretation that the animal is simply searhing elsewhere for food rather than speifially trying to esape aversive ues. At a physiologial level, the marked NCEs ourring on the first day are not aompanied by the expeted inreases in stress as measured by irulating ortiosterone (Flaherty et al., 1985). Furthermore, the administration of anxiolyti drugs on the first postshift day does not reliably redue the size of the NCE (Flaherty, 1991). Even the inrease in ortiosterone observed on the seond postshift day may not uniquely support a frustration or stress aount. Elevated ortiosterone levels are also related to hanges in energy metabolism (Miller & Tyrrell, 1995), espeially during fasting (Owen, 1989). The seond most ommon explanation for NCE is based on the role of assoiative generalization derements, beause of hanges in the sensory-pereptual-motivational ues related to reward (Capaldi, 1972; Spear & Spitzner, 1966). This aount is also inomplete, suffering from the same problems of response speifiation and alternative interpretations noted above for the frustration aount. In addition, the generalization derement aount predits NCE should our from any modifiation of the stimulus onditions, inluding upshifts in the amount of food or the 153
154 PECORARO, TIMBERLAKE, AND TINSLEY introdution of a novel tone after the shift. Neither of these preditions is supported (Lombardi & Flaherty, 1978). The purpose of the present experiments was to examine more arefully the funtional searh hypothesis of NCE first advaned by Elliott (1928). He argued that the inrease in erroneous entries into unrewarded arms of a maze, following downshifts in reward, was due to the rat's searh for the previously experiened reward. In a related vein, Flaherty (1996) reently proposed a multistage, "searh-thenemotion" hypothesis to aount for data showing an inrease in loomotion and open rearing in an arena, following a downshift in surose onentration (Flaherty, Blitzer, & Collier, 1978; Flaherty, Powell, & Hamilton, 1979). The basi searh hypothesis an be amplified by larifying the basis of searh and the form it takes. In a behavior systems's view, the results of laboratory manipulations using food rewards are based on the foraging mehanisms and strategies harateristi of the speies under investigation (Silva & Timberlake, 1997; Timberlake & Luas, 1989). In this view, a marked redution in available food disrupts an animal's urrent or antiipated energy flow and moves the animal from a onsummatory motivational mode that supports handling or ingestive behaviors to foal and general searh modes that under more natural irumstanes would failitate regaining the missing reward, thus dereasing intake and inreasing searh. Eah searh mode primes a repertoire of speifi stimulus sensitivities and searh responses potentially effetive in dealing with natural foraging problems and may be onditioned to various ues (e.g., Silva & Timberlake, 1997; Silva, Timberlake, & Gont, 1998). General searh modes should potentiate loomotor searh and general orienting responses aimed at loating alternative soures of food, whereas foal searh modes should inrease loal searh for and sampling of alternative food soures, resembling arearestrited searh disussed in ethologial literature (e.g., Bell, 1991). The searh hypothesis by itself predits no inrease in searh behaviors with upshifts, and no partiular avoidane of or esape from downshifted rewards. If anything, foal searh modes should be onditioned most strongly to areas prediting food, thus generating more speifi investigation of these areas. The three experiments reported here sought to evaluate experimental evidene for the funtional hypothesis that downshifts in the attrativeness of surose solutions should evoke searh modes and ativate repertoires of searh behavior. The apparatus was a four-arm radial maze with a surose solution at the end of one or more arms. The first experiment ompared funtional and frustrative hypotheses, onerning the level and omprehensiveness of searh behaviors, the aversiveness of the downshifted arm, and the role of presumed frustration during training in inreasing persistene. The seond experiment ompared the effets of reward downshifts and-upshifts on the evoation of searh repertoires. Finally, the third experiment questioned whether the repertoire of searh responses produed by inentive downshifts depended on the availability of multiple alternative solutions introdued during the shift. Experiment 1 The purpose of Experiment 1 was to examine the relation of funtional searh behaviors and NCEs by observing the behavior of rats shifted from a 32% to a 4% surose solution with that of a ontrol group always reeiving 4% surose. During training, all platforms ontained beakers, but the inentive solution was put in only one beaker at the same, fixed loation in a four-arm radial maze. Following the downshift, a 4% solution was plaed in all four arms. The shifted group finding the preshift solution in the same fixed loation was referred to as 32 FL-4, whereas the ontrol group was referred to as 4 FL-4. We also inluded a variable-loation shifted group that originally found their 32% preshift solution randomly in one of the four arms eah trial (Group 32 VL-4). Beginning with the shift day, this group also was presented with a 4% solution at the end of eah arm. This experiment was inspired by the disrepant results of Flaherty et al. (1978) and Flaherty et al. (1979), who tested the effets of shifting rats from 32% to 4% surose in an open field on the shift day that ontained 4% bottles at four separate loations. In both experiments, ambulation and rearing inreased in the shifted rats as ompared with unshifted rats, whereas inreased sampling from the four bottles ourred only in the seond experiment. If one uses the rats' systemati sampling behavior to distinguish emotional energization from searh hypotheses, the first experiment (Flaherty et al., 1978) supported the frustration hypothesis, whereas the seond experiment (Flaherty et al., 1979) supported the searh hypothesis. Although Flaherty et al. (1979) suggested that the rats may have learned the devaluation with minimal inreases in sampling, this would not explain the apparent failure of many shifted rats to sample several of the novel tubes in the first experiment. On the other hand, the rats' rapid assessment of the solution onentration ombined with the experimental proedure of oding behavior only one every 10 s ould aount for the differenes in results. To avoid missing sampling bouts shorter than 10 s, we oded behaviors in 1-s bins. To avoid onfusing sampling with ingestion, we ounted only the first bout of beaker sampling following eah platform entry. Our design allowed us to test a number of hypotheses. First, if the downshift from 32% to 4% surose evokes a repertoire of general and foal searh behaviors related to gathering food, then the result should be inreased orienting, nose-down loomotion, entries to all loations, and systemati sampling of novel solutions for all shifted groups. The fixed training group may show a plae preferene for the training platform in the form of area-restrited searh, beause in the past that platform best predited onsumption of 32% surose, whereas the variable-loation group should form no plae preferenes. In ontrast, if the inentive downshift produes general frustration, all shifted groups should show emotional behaviors, inluding withdrawal from any area of redued reward, energization of ongoing behaviors, unsystemati sampling
SEARCH AND INCENTIVE DOWNSHIFTS 155 of loations and solutions, and possibly, displaement grooming. Also, beause the variable-loation group should have experiened the frustration of not finding reward in the first arm hosen on most training days, frustration should beome onditioned to eventually finding surose. Therefore, Group 32 VL-4 should show relatively enhaned instrumental searh, following a downshift than Group 32 FL-4. Subjets Method Subjets were thirty naive, 90-day-old female Sprague-Dawley rats (Rattus norvegius) bred in the department olony. Animals were housed separately, water was available ad libitum, and the olony lighting shedule was a 12-hr light-dark yle, lights on and off at 0600 and 1800, respetively. The experiments were run approximately between 1200 and 1600. Animals were maintained at 85% free-feeding weight by one daily feedings of rat how in the olony given 4 5 hr following the ingestion of surose in the experimental situation. Animals that drank no more than 4 ml per day of a surose solution within the first 10 days of the preshift phase were eliminated from the experiment. Apparatus and Materials The rats were fed surose solutions on an elevated plus maze (1.67 m in irumferene, having a 32.5-m enter platform irumferene, 30.5- X 23-m end platforms, 11.5-m width arms, and 66-m off ground) loated in a 2.13-m 2 experimental room. A Panasoni BL200 video amera was loated 1.5 m above the maze. The room was dimly lighted by an inandesent lamp. The solutions, mixed by weight from ane sugar and tap water every several days, were presented to the animals in opaque plasti beakers. Proedure The experiment was divided into a preshift phase and a postshift phase. During the preshift phase, whih lasted for 14 days, two groups of rats were fed a 32% surose solution that was loated in a beaker (a 4.7-m X 2.8-m diameter, 40- polyethylene up) at the end of one of the arms of the maze. One of these groups reeived the surose solution in the same fixed loation eah day and is designated 32 EL-4. The other group reeived the surose solution in a single loation that was varied eah day and is designated 32 VL-4. During the postshift phase, whih lasted for 4 days, these two groups were shifted to a 4% surose solution that was available on all four platforms of the maze. The ontrol group, 4 FL-4, reeived a 4% surose solution in a fixed loation during training and then reeived the same 4% solution in all arms during the postshift. Groups were mathed by weight and then randomly assigned to ondition. The rats were transported from the olony in small arrying ases 4 at a time. They were plaed 1 at a tune in a holding pen on the enter of the maze for several seonds, and then they were released and allowed to move freely for 5 min, while the experimenter monitored them on video from another room. A beaker of surose solution, either a 4% or 32% surose solution, was available in one of the arms. All rats were given more of a solution than they ould onsume in the 5-min period. All other arms ontained empty beakers during the preshift phase. All beakers were visible from the maze surfae. Following a shift, all four beakers ontained solutions exept for one group in Experiment 3. The maze was leaned between animals with onentrated Roal-D (Winthrop Veterinary, New York, NY), a detergent, viruide, and deodorant (diluted 400:1). Dependent Measures and Coding Three main dependent measures were monitored: The daily intake of surose solutions, five types of behavior, and nine spatial loations. Both the behaviors and loations of the rats were oded separately. Coders wathed real-time videos and ontinuously depressed the key on the omputer keyboard symboli of the behavior or loation urrently being observed. A omputer program reorded the keystroke urrently depressed at the edge of eah 1-s bin. Additional ontinuous depression of the key within the 1-s bin had no reorded effet. Thus, the omputer was, in effet, sampling the behavior of the oder. Videotapes of several key sessions, namely, the 2 days prior to a shift in reward and the 2 days following a shift in reward were oded. Fifteen perent of tapes from eah type of oding session (e.g., behaviors or loations) and eah type of day (e.g., preshift, postshift, et.) were randomly seleted to be reeded by an independent observer, unaware of the situation. Reliability was alulated by dividing the number of agreements per 1-s bin by the number of total 1-s bins in a session. Reliability was 97% for loations and 85% for behaviors. All behaviors and loations were oded ategorially. Desriptions of the oding shemes are given below. Loations. The maze was divided into nine zones: the enter platform, the four arms, and the four feeding platforms, inluding the target platform (TP), where the high-quality surose was loated, (hi Experiment 1 in whih surose loations were variable for one group, the target platform refers to the loation where surose was last found by the rats.) Nontarget platforms were designated in terms of their lokwise (CW) relation to the target platform (i.e., CW1, CW2, and CW3). Zone entries were sored when an animal extended approximately half of a body length into a zone. Beaker orientation. This ategory primarily inluded drinking from or being oriented to a beaker, and inluded the oasional pauses in protrated drinking bouts as well. The animal's nose had to be in the beaker, just above it, or investigating the exterior of the beaker. There were beakers in all arms throughout the experiment. (Beause the time spent in beaker orientation was the approximate inverse of time spent in nose-down loomotion in every experiment, only results for nose-down loomotion are reported, and engagement in beaker orientation is implied. In all of the experiments to be reported, nose-down loomotion and beaker orientation together aounted for at least 80% of all behavioral time alloation.) Nose-down loomotion. The key harateristi of nose-down loomotion was that the rat's nose atively explored the surfae or edges of the maze without breaking the horizontal plane or being in a beaker ontaining solution. Generally, this behavior inluded ontinuous ambulation with oasional pauses while sniffing ontinued. Loomotion without sniffing was very rare and was inluded as a ase of nose-down loomotion. Nose-up posture. The nose-up posture was sored when the animal's nose rose, without the animal rearing, above an imaginary horizontal plane parallel to the plane of the maze surfae and entered on the apex of the rat's nose in the sagittal view. During nose-up behavior, the rat tended to ease ambulation, and it strethed and sniffed the air. This behavior often preeded open rearing.
156 PECORARO, TIMBERLAKE, AND TINSLEY Open rearing. Open rearing inluded all episodes of behavior when the rat's forepaws left the maze floor and the animal supported itself on its hind legs only, the animal apparently sampling the air and room ues, as indiated by sniffing and turning of the head. Beaker sampling. Beaker sampling was designated as the first episode of beaker orientation following an entry to a feeding platform. Thus, any single entry to a platform ould allow multiple bouts of beaker orientation, but only the first was ounted as a ase of sampling. Reording a seond bout of sampling on that same platform required that the animal leave the platform and then return to it for another bout of beaker orientation. Grooming. Grooming inluded episodes of srathing with fore- or hind-legs, washing the fae, or liking the paws. Beause grooming behavior did not vary systematially as a funtion of group or day of shift, and aounted for minimal total behavioral time, it was exluded from this report. Statistis. All analyses were performed on 2-day averages of raw data, exept in the ase of beaker sampling, whih was oded on the last day of the preshift and the first day of eah postshift. Analyses of variane (ANOVAs) were used to determine the statistial signifiane of dependent measures. One animal from 32 FL-4 was exluded for failure to meet predetermined preshiftintake riteria. Intake Results Preshift Downshift Preshift Downshift H 32 FL-4 EJ 32 VL-4 4 FL-4 o> (0 80i To demonstrate the basi phenomenon of onsummatory ontrast, intake data were ombined into 2-day bloks and ompared for eah group aross preshift and postshift days. Figure 1A shows mean surose onsumption averaged over 2-day bloks for all groups aross eah day type. A one-way ANOVA omparing preshift intake showed no differenes in intake between groups, F(2, 26) = 1.02, p >.05, whereas a one-way ANOVA omparing postshift onsumption be- 60- o '+* us o Preshift Downshift Preshift Downshift Figure 1. (A) Mean intake of surose solutions in ml (±SEM) aross suessive days for shifted and unshifted groups (Experiment 1). (B) Mean seonds (±SEM) engaged in nose-down loomotion for eah group during the preshift and the downshift in 2-day bloks (Experiment 1). (C) Mean number of entries to new loations (±SEM) for eah group during the preshift and downshift in two day bloks (Experiment 1). (D) Mean number of bouts of general orienting responses (i.e., nose-up posture and open rearing) for eah group (±SEM) during the preshift and downshift in 2-day bloks (Experiment 1). 32 FL-4 = fixed-loation training, shifted; 32 VL-4 = variable-loation training, shifted; 4 FL-4 = fixed-loation training, unshifted.
SEARCH AND INCENTIVE DOWNSHIFTS 157 tween groups was signifiant, F(2, 26) = 48.59, p <.001. Planned omparisons showed that both 32 FL-4 and 32 VL-4 drank less of the postshift solution than the ontrol group, F(l, 26) = 60.04,;? <.001, andf(l, 26) = 82.71,p <.001, thus demonstrating a robust negative onsummatory ontrast effet. Nose-Down Loomotion To test the hypothesis that appetitive behaviors would inrease when shifted animals suppressed their intake, groups were ompared with respet to time engaged in nose-down loomotion. Figure IB shows 2-day averages of time alloated to nose-down loomotion as a funtion of shift day for eah group. A one-way ANOVA omparing preshift levels of nose-down loomotion between groups was signifiant, F(2, 26) = 8.44, p <.002. Planned omparisons showed no differenes in loomotion between 32 FL-4 and 4 FL-4, F(l, 26) = 1.71, p >.05, whereas 32 VL-4 expressed more nose-down loomotion than 32 FL-4, F(l, 26) = 16.01, p <.001, and 4 FL-4, F(l, 26) = 7.63, p <.02. This additional loomotion by 32 VL-4 was expeted beause the surose solution was randomly loated on the maze. In ontrast to this proedure-generated preshift differene in loomotion, the downshift produed large differenes between groups that were not neessitated by training. A one-way ANOVA omparing nose-down loomotion between groups was signifiant, F(2, 26) = 54.94, p <.001. Planned omparisons showed that both 32 FL-4 and 32 VL-4 engaged in far more nose-down loomotion than the unshifted group, F(l, 26) = 54.94, p <.001, and F(l, 26) = 52.59, p <.001, respetively. As expeted, in ontrast to the onsummatory response suppression resulting from the downshift, appetitive behaviors were enormously failitated in shifted groups. A planned omparison showed that this failitation was slightly greater in the shifted group trained in various loations than in the shifted group trained in a fixed loation, F(l, 26) = 6.86, p <.02, a predition extrapolated from frustration theory. Loation Changes To show that animals were in fat traversing the maze during their inrease in ambulation, omparisons were made of the number of atual loations hanges for eah group. Figure 1C shows 2-day averages of the number of loations hanges made by eah group as a funtion of shift day. Although a one-way ANOVA showed differenes in loation hanges between groups during training, F(2, 26) = 3.88, p <.05, whih again is probably a neessary result of the different training proedures, planned omparisons showed no differenes in the number of loation hanges between 32 FL-4 and 4 FL-4, F(l, 26) < 1.0, whereas 32 VL-4 made somewhat more loation hanges than both 32 FL-4, F(l, 26) = 5.97, p <.05, and 4 FL-4, F(l, 26) = 5.54, p <.05. More importantly, a one-way ANOVA omparing the number of loation hanges during the downshift was also signifiant, F(2, 26) = 48.32, p <.001. Planned omparisons showed that both 32 FL-4 and 32 VL-4 made many more loation hanges following the downshift than did the ontrol group, F(l, 26) = 80.84, p <.001, and F(l, 26) = 61.60, p <.001, without showing differenes between themselves, F(l, 26) = 1.82, p >.05. Thus, shifted animals were, in fat, traversing the maze muh more than unshifted animals during the downshift. Sampling data presented later will reveal that these traversals were systemati. Orienting Responses Beause it is possible that inreases in loation hanges and loomotion might represent inreases in general drive alone beause of frustration, we also tested differenes in orienting behaviors between groups, beause suh behaviors are likely to be funtionally related to searh. Figure ID shows 2-day averages of hanges in the number of bouts of open rearing and nose-up postures for eah group on preand postshift days. A one-way ANOVA showed no differenes in orienting behaviors between groups prior to the shift, F(2, 26) < 1.0, whereas signifiant differenes existed between groups following the shift, F(2, 26) = 14.66, p <.001. Planned omparisons revealed no differenes between the shifted groups, F(l, 26) = 1.62, p >.05, whereas both 32 FL-4 and 32 VL-4 expressed more bouts of orienting than the ontrol group, F(l, 26) = 14.47, p <.001, and F(l, 26) = 27.21, p <.001, respetively. Beaker Sampling To test whether the apparent searh behavior was direted speifially toward food goals, groups were ompared for levels of sampling behaviors toward both new and old food loations both before and after the shift. Figure 2 shows hanges in the sampling behavior of eah group as a funtion of shift day. The left panel shows total mean sampling on all platforms during the last day preshift, whereas the right panel shows sampling on eah platform on the first day of the downshift. A one-way ANOVA showed no differenes in 3 O 1 D) C "5. ro a> ro ID Preshift CW1 CW2 CW3 Downshift by platform Figure 2. Mean number of sampling bouts (±SEM) during the last-day preshift (left panel) and the mean number of sampling bouts per platform during the first day of the downshift (right panel). 32 FL-4 = fixed-loation training, shifted; 32 VL-4 = variable-loation training, shifted; 4 FL-4 = fixed-loation training, unshifted; TP = target platform; CW = lokwise.
158 PECORARO, TIMBERLAKE, AND TINSLEY sampling bouts between groups during the preshift, F(2, 26) < 1.0. In ontrast, a one-way ANOVA of sampling during the downshift showed signifiant differenes between groups, F(2, 26) = 39.70, p <.001. Planned omparisons showed that both 32 FL-4 and 32 VL-4 showed greater sampling than the ontrol group, F(l, 26) = 78.35,p <.001, and F(l, 26) = 28.21; p <.001, respetively. Consistent with Flaherty et al. (1979) and ontrary to the findings of Flaherty et al. (1978), sampling of novel beakers inreased following the shift. In addition, a planned omparison also showed greater postshift sampling in Group 32 FL-4 than in Group 32 VL-4, F(l, 26) = 12.54, p <.002, indiating that variable loation training had no failitatory effet on this instrumental behavior. Finally, in order to test the frustration hypothesis that the target platform should beome relatively aversive following the shift versus the searh hypothesis that the platform should be relatively preferred, the distributions of eah group's sampling on various platforms were ompared. A 3X4 (Group X Platform) ANOVA was performed for the first day of the downshift, revealing main effets of group, F(2,26) = 39.70, ;> <.001, and platform, F(3, 78) = 10.22, p <.001, and a Group X Platform interation, F(6, 78) = 6.67, p <.001. Planned omparisons showed greater sampling by 32 FL-4 than by 32 VL-4, F(l, 26) = 12.54, p <.002, and greater sampling by 32 VL-4 than by 4 FL-4, F(l, 26) = 28.21,p <.001. No differenes in sampling were found between 32 FL-4 and 32 VL-4 on nontarget platforms, F(l, 26) < 1.0, whereas 32 FL-4 showed far more sampling bouts than 32 VL-4 on the target platform, F(l, 26) = 29.05, p <.001, a finding that (a) suggests the formation of a plae preferene as a result of fixed loation training, and (b) runs ounter to the predition of the frustration hypothesis that the persistene of searh should be onditioned to frustrative ues. Disussion A repertoire of searh behaviors robustly aompanied the negative onsummatory ontrast in shifted groups relative to the ontrols. Relative to unshifted ontrols, both shifted groups showed dramati inreases in nose-down loomotion, general orienting responses, entries to loations on the maze, and sampling of novel beakers. Seonds of nose-down loomotion and beaker : orientation were inversely related and aounted for over 80% of behavioral time in all phases of the experiment. The unshifted group that reeived a 4% solution all along, on the other hand, showed no NCEs and no signifiant realloations of behavior. Our finding that shifted rats sampled from novel solutions was robust. Shifted animals had a muh greater propensity to enter other solution-ontaining platforms and sample the beakers following the shift than did unshifted animals. The main differene between this study and that of Flaherty et al. (1978) seems to be the frequeny with whih investigators oded behaviors. Flaherty et al. (1978) oded behaviors one per 10 s, whereas rats hi this study were oded one per seond. Our sampling riterion, only the first instane of beaker orientation (drinking or sniffing) following an entry to a platform, would appear to be a onservative estimate of total sampling, and thus, a true and robust onsequene of the inentive downshift. Several points of omparison between the two shifted groups are also worth mentioning with respet to frustration theory. Although Group 32 VL-4 did show slightly more nose-down loomotion than did the Group 32 FL-4 following the shift, they showed the same results prior to the shift as well. Further, there were no differenes between these groups in the number of total loation hanges, or in number of general orienting responses. Also, the group trained in a fixed loation was atually more persistent in terms of overall sampling than was the group trained in varied loations. The majority of this additional sampling onsisted of additional entries to the fixed loation in whih they were trained, apparently beause of the plae preferene engendered by fixed-loation training. Entries to the other platforms were equivalent. The failure to find enhaned loomotion and sampling as a result of training in variable loations argues against the importane of onditioned frustration for the variable group, whose frustration presumably should have been onditioned to environmental or proprioeptive ues prediting reward. Finally, the plae preferene shown by 32 FL-4 for the platform formerly ontaining 32% surose argues strongly against the hypothesis that the lous of redued reward beomes relatively aversive following the downshift. It rather seems that the animals superimposed their training habits on their subsequent searh behavior, not an unreasonable average hoie in a natural foraging situation. In a related finding, Meler and Timberlake (1985) showed that water-deprived rats ontinued visiting a preferred arm no longer ontaining water on 25% of their visits. Moreover, even when an arm ontaining saharin solutions was made expliitly aversive by pairing that arm with a lithium hloride injetion, the rats ontinued to visit that arm on subsequent days (Meler & Timberlake, 1985). Thus, the overall impression formed from the urrent data is that the predominant effet of an inentive downshift was funtional and systemati searh behavior rather than inreased emotional behavior, random energization, or a speifi assoiatively based avoidane of the downshifted loation. Experiment 2 The primary purpose of Experiment 2 was to determine whether the behavioral hanges observed in Experiment 1 were part of a speifi foraging repertoire evoked by a redution in the austomed inentive quality or part of a repertoire of exploratory or emotional responses simply evoked by any inentive hange. To test this, we ompared a group of animals reeiving 4% surose with a group that was shifted from 4% to 32% and with a group that ontinued to reeive 32% surose. If we are orret that NCE our in the ontext of an evoked funtional searh mode, then we would expet an inentive upshift to produe neither ontrast effets or inreases in loomotor and sampling behaviors. Suh a finding in the present apparatus would also provide
SEARCH AND INCENTIVE DOWNSHIFTS 159 additional evidene that NCEs do not result from the operation of a symmetrial exitatory generalization gradient. Method Subjets, Apparatus, Materials, Dependent Measures, and Coding The types of subjets, the apparatus, materials, dependent measures, oding methods, and statistial analyses were the same in Experiment 2 as in Experiment 1. The proedure was the same exept where noted. One animal was eliminated from Group 32-32-4 for failure to drink. Proedure Three groups of 10 animals eah reeived three phases of training. The main experimental group reeived 4% surose for 20 days in Phase 1, was briefly upshifted to a 32% surose for 4 days in Phase 2, and was then downshifted to a 4% surose solution for 2 days in Phase 3. On the basis of the order of solution presentation, this upshifted group was designated as 4-32-4. The group ontrolling for negative ontrast reeived 4% surose in all three phases, and it was designated as 4-4-4. The group ontrolling for positive ontrast in Phase 2 reeived 32% surose in both Phases 1 and 2, was then downshifted to a 4% solution in Phase 3, and was designated as 32-32-4. Group 32-32-4 also reeived the most training with surose prior to the downshift (20 days more than 4-32-4). Exept for sampling data, whih onsisted of data from the first day of the downshift, all other data reported onsisted of 2-day averages: the 2 days prior to the upshift, the first 2 days of the upshift, and the first 2 days of the downshift. One animal from Group 32-32-4 was eliminated for failing to meet intake riteria. Intake Results To test whether positive or negative ontrast ourred following upshifts and downshifts, respetively, betweengroup ANOVAs were onduted on 2-day averages of intake data from preshift, upshift, and downshift days. It was expeted that neither positive or negative onsummatory ontrast would our during the upshift, whereas negative ontrast would result from the downshift. Figure 3A shows mean surose onsumption averaged over 2-day bloks for all groups aross eah day type. A one-way ANOVA omparing preshift intake levels showed no differenes between groups, F(2, 26) = 2.46, p >.05. A one-way ANOVA omparing intake levels during the upshift, however, was signifiant, F(2, 26) = 4.49, p <.05. Planned omparisons showed that (a) both 32-32-4 and 4-32-4 drank more than the unshifted ontrol group (4-4-4) during the upshift, F(l, 26) = 8.14,p <.01, andf(l, 26) = 4.78, p <.05, respetively, but that the shifted groups did not differ amongst themselves, F(l, 26) < 1.0, indiating no evidene for positive ontrast during the upshift. The differene between the shifted and unshifted groups on preshift and upshift days seems to be a ase of ontinuing aquisition for the shifted groups at a time when the ontrol group had reahed asymptote. Finally, a one-way ANOVA omparing downshift intake levels was signifiant, F(2, 26) = 12.35, p <.001. Planned omparisons showed that 4-32-4 showed a trend to drink less than 4-4-4, F(l, 26) = 3.5, p =.07, and 32-32-4 drank signifiantly less than 4-32-4, F(l, 26) = 9.68, p <.005. The mere trend toward negative ontrast by Group 4-32-4 appears to be due to a fast reovery on the seond day of the downshift, whereas the group having more protrated experiene with the 32% solution showed more persistent negative ontrast. Nose-Down Loomotion To test our predition that searh behaviors would aompany downshifts but not upshifts, eah group's loomotor levels were ompared during the preshift, upshift, and downshift. Figure 3B shows 2-day averages of seonds of nose-down loomotion for eah group during preshift, upshift, and downshift days. One-way ANOVAs omparing levels of loomotion between groups onfirmed our predition of no differenes between groups during the preshift, F(2, 26) = 2.73, p >.05, or during the upshift, F(2, 26) = 1.64, p >.05, whereas following the downshift, signifiant differenes in loomotion between groups existed, F(2, 26) = 12.33, p <.001. Planned omparisons showed that 32-32-4 spent more time engaged in nose-down loomotion than 4-32-4, F(l, 26) = 5.15,p <.05, and 4-32-4 engaged in more loomotion than 4-4-4, F(l, 26) = 7.58, p <.02. As expeted, inreases in nose-down loomotion ourred for shifted groups during the downshift, but not during the upshift. Loation Changes To show that shifted groups were, in fat, traversing the maze more than the unshifted group, loation hanges between groups were ompared aross day types. Figure 3C shows 2-day averages of loation hanges aross days for eah group. One-way ANOVAs showed no differenes in loation hanges between groups during the preshift, F(2, 26) = 2.56, p >.05, and during the upshift, F(2,26) = 1.63, p >.05, whereas during the downshift, signifiant differenes existed between groups, F(2, 26) = 19.44, p <.001. Planned omparisons showed that 32-32-4 traversed the maze more than 4-32-4, F(l, 26) = 10.54, p <.001, and 4-32-4 traversed the maze more than 4-4-4, F(l, 26) = 9.42, p <.005. Thus, signifiant maze traversals aompanied loomotor inreases. Orienting Responses Beause loomotor responses might represent inreases in general drive alone beause of frustration, we tested differenes in orienting behaviors between groups, insofar as suh behaviors may indiate searh. Figure 3D shows 2-day averages of hanges in the number of bouts of open rearing and nose-up postures for eah group on preshift and shift days. One-way ANOVAs showed no differenes in orienting responses between groups during the preshift, F(2, 26) < 1.0, or during the upshift, F(2, 26) < 1.0. During the downshift, however, signifiant differenes in orienting
160 PECORARO, TIMBERLAKE, AND TINSLEY en CO O C o a o Figure 3. (A) Mean intake of surose solutions (in ml; ±SEM) aross suessive days for shifted and unshifted groups (Experiment 2). (B) Mean seonds engaged in nose-down loomotion (±SEM) for eah group during the preshift, the upshift, and the downshift in 2-day bloks (Experiment 2). (C) Mean number of entries to new loations (±SEM) for eah group during the preshift, the upshift, and the downshift in 2-day bloks (Experiment 2). (D) Mean number of bouts of general orienting responses (i.e., nose-up posture and open rearing) for eah group (±SEM) during the preshift, the upshift, and the downshift in 2-day bloks (Experiment 2). 32-32-4 = downshifted only; 4-32-4 = upshifted and downshifted; 4-4-4 = unshifted. responses between groups existed, F(2,26) = 4.91, p <.02. Group 32-32-4 made more orienting responses than did the unshifted group during the downshift, F(l, 26) = 9.72, p <.005, whereas 4-32-4 did not, F(l, 26) = 1.52, p >.05. Planned omparisons also showed a trend for greater orienting responses by 32-32-4 than did 4-32-4, F(l, 26) = 3.68, p =.06. Thus, it one again appears that signifiant experiene with a preferred food results in enhaned orienting behaviors during a subsequent food redution. Beaker Sampling To distinguish between the general ativating properties of frustration versus the diretedness of searh behaviors, as well as to show that inreased searh was harateristi of downshifts but not of upshifts, the sampling of beakers in new loations was ompared between groups, between the preshift and the upshift, and between the upshift and the downshift. Figure 4 shows sampling behavior for eah group on the last day of the upshift and first day of the downshift. The left panel shows total mean bouts of beaker sampling during the upshift, and the right panel shows mean bouts of beaker sampling on eah platform for eah group. With respet to differenes in total bouts of sampling on upshift and downshift days, a 3 X 2 (Group X Day) ANOVA showed main effets of group, F(2, 26) = 15.82, p <.001, and day, F(l, 26) = 70.13, p <.001, and a Group X Day
SEARCH AND INCENTIVE DOWNSHIFTS 161 a) + * 3 O O) D. E to to n o> 5 32-32-4 0 4-32-4 4-4-4 Upshift TP CW1 CW2 CW3 Downshift by platform Figure 4. Mean number of sampling bouts (±SEM) during the last day of the upshift (left panel) and the mean number of sampling bouts per platform during the first day of the downshift (right panel). 32-32-4 = downshifted only; 4-32-4 = upshifted and downshifted; 4-4-4 = unshifted; TP = target platform; CW = lokwise. interation, F(2,26) = 14.18, p <.001. Aone-way ANOVA of groups during the upshift showed no differenes in total sampling, F(2,26) = 1.10,p >.05. Planned omparisons of upshift and downshift days showed no signifiant inrease in sampling during the downshift for Group 4-4-4, F(l, 26) = 1.07, p >.05, whereas both 32-32-4 and 4-32-4 showed signifiant inreases in sampling during the downshift, F(l, 26) = 69.10, p <.001, and F(l, 26) = 24.76, p <.001, respetively. 32-32-4 also showed more sampling than 4-32-4 during the downshift, F(l, 26) = 5.26, p <.05. As expeted, sampling behavior did not inrease during the upshift, whereas it did inrease during the downshift, along with the other searh-like behaviors. Finally, to test whether the downshifted platform beame aversive, we ompared the distribution of sampling on various platforms during the downshift. A 3 X 4 (Group X Platform) ANOVA on sampling during the downshift showed main effets of group, F(2, 26) = 20.76, p <.001, greater sampling on the target platform than other platforms, F(3, 78) = 11.73, p <.001, and a Group X Day interation, F(6, 78) = 2.54,^ <.05. Planned omparisons showed that only Group 32-32-4 sampled from the target platform signifiantly more than the average other platform, F(l, 8) = 37.50, p <.001, onsistent with the view that the training resulted in a plae preferene that persisted during the downshift. Disussion The main finding of Experiment 2 is that no NCEs or major realloation of behavior aompanied upshifts in reward, whereas the downshift alled forth robust NCEs and large-sale realloations of behavior in the same rats. Although positive ontrast effets (PCE) have oasionally been reported following inentive upshifts, usually following some prior experiene with food redution, they were not found here. Though failure of positive ontrast may have been due to possible eiling effets, our observations, along with all previous failures to find NCE following upshifts, ontinue to pose a grave problem for a generi generalization hypothesis of NCE. Furthermore, generalization hypotheses do not predit the type of hange in appetitive behavior seen here. The behavioral hanges following the downshift in this experiment losely paralleled those found in the previous experiment: Shifted rats showed a large suppression of intake ombined with inreased nose-down loomotion, loation entries, orienting postures, and sampling relative to the unshifted group, in addition to an apparent preferene, rather than aversion for the downshifted inentive loations. Experiment 3 The primary purpose of Experiment 3 was to determine the extent to whih the NCE and inreased searh behaviors, following inentive downshifts in Experiments 1 and 2, depended on the plaement of 4% surose solutions in all the arms on the shift day. It might be argued that the searh behaviors were not so muh evoked by the inentive downshift itself as they were eliited and supported by the disovery of alternative soures of surose (albeit of a less-preferred onentration). Arguing against this interpretation is the fat that no large-sale inrease in searh ourred for unshifted ontrols when the 4% solution was introdued to all four platforms. Furthermore, the suddenness and regularity of inreased searh displayed by shifted groups in the two previous experiments suggested strongly evoked omponents. To this point, our hypothesis has been that searh behaviors are alled forth and supported by a swith in motivational modes from onsummatory to searh modes. Our assumption has been that these hanges in behavior have been evoked by the inentive downshift and would our whether or not there are multiple loations of the downshifted surose solution. To test this assumption, we ompared two groups of rats downshifted from 32% to 4% surose solutions, one group presented with 4% solutions in all arms, one with a 4% solution only in the trained arm. The unshifted ontrol group also reeived a 4% solution only in a single arm. Our predition was that the inentive downshift itself would be suffiient to evoke systemati searh, regardless of the number of alternative solutions. A seond purpose of Experiment 3 was to test the effets of an inentive upshift on these searh behaviors one they have been evoked. If inreases in loomotion, orienting, entries, and sampling during a downshift represent searh behavior for the missing ommodity, then returning the missing ommodity should halt these behaviors immediately. However, if an upshift to a 32% solution produed a positive hedoni state, for example, an "elation" effet (see Crespi, 1942) relative to the 4% solution experiened during the downshift, the inentive upshift ould be aompanied by a positive ontrast effet in appetitive behaviors, suh as an inrease in loomotor ativation as has been reported in runway situations following an upshift (Benefield, Osos, &
162 PECORARO, TIMBERLAKE, AND TINSLEY Ehrenfreund, 1974; Crespi, 1942; Ehrenfreund & Badia, 1962) or on a maze following an upshift (Shanab & Ferrell, 1970). We tested these possibilities by reversing the order of shifts used in Experiment 2: In Experiment 3, a downshift to a 4% solution was followed by an upshift to a 32% solution. Beause the animals.will have experiened the higher onentration prior to the upshift, any derements in positive ontrast effets due to generalization should be minimized. Method Subjets, Apparatus, Materials, Dependent Measures, and Coding The types of subjets, apparatus, materials, dependent measures, and oding were the same in Experiment 3 as in Experiments 1 and 2. One animal was eliminated from Group 32-4'-32 for failing to drink. Proedure Experiment 3 was similar to Experiment 2 in that it used 3 groups of 10 rats in three experimental phases. It differed mainly in reversing the order of the upshift and the downshift. The ontrol group, 4-4'-4, reeived a 4% solution in a fixed arm of the maze during all three phases. Two shifted groups reeived a 32% solution in a fixed arm of the maze for 29 days, were shifted to a 4% solution for 2 days, and were then upshifted to a 32% solution for 2 more days. In Phase 1, all animals reeived one solution in a fixed loation, with the other arms being devoid of both beakers and solutions. During the downshift, Groups 4-4'-4 and 32-4'-32 reeived a single 4% solution in the same arm of the maze, with other platforms ontinuing to be devoid of both beakers and solutions, whereas Group 32-4 4-32 reeived a single 4% solution in all four arms of the maze during the downshift. During the subsequent upshift, eah group ontinued reeiving the same number of solutions it had reeived during the downshift. Intake Results To demonstrate negative onsummatory ontrast, we ombined intake data for eah group into 2-day bloks for eah day type (i.e., preshift, downshift, and upshift days), and the groups were ompared. As in the previous experiment, negative ontrast was expeted only during the downshift. Figure 5A shows mean surose onsumption averaged over 2-day bloks for all groups aross eah day type. A one-way ANOVA omparing intake levels between groups during the preshift showed a trend for the shifted groups to drink more than the ontrol group, but no signifiant differenes, F(2, 26) = 2.92, p >.05. A one-way ANOVA of downshift intake levels, however, showed signifiant differenes between groups, F(2, 26) = 16.58, p <.001. As expeted, planned omparisons showed that both Groups 32-4'-32 and 32-4 4-32 onsiderably suppressed their intake relative to the ontrol group, F(l, 26) = 32.78, p <.001, and F(l, 26) = 11.08, p <.005, respetively, thus demonstrating negative onsummatory ontrast. Group 32-4 J -32 drank slightly less than Group 32-4 4-32 during the downshift, F(l, 26) = 6.18, p <.02. A one-way ANOVA omparing intake levels during the upshift was also signifiant, F(2, 26) = 6.69, p <.005. Planned omparisons showed that both 32-4'-32 and 32-4 4-32 drank more than the ontrol group, F(l, 26) = 6.79, p <.02, and F(l, 26) = 12.33, p <.002, essentially returning to the preshift drinking levels. Nose-Down Loomotion To show that the large-sale inreases in loomotion were evoked by the downshift but not by the upshift, groups were ompared for differenes in loomotion during the preshift, the downshift, and the upshift. Figure 5B shows 2-day averages of hanges in seonds of nose-down loomotion during preshift, the downshift, and the upshift for eah group. A one-way ANOVA omparing preshift levels of loomotion revealed no differenes between groups, F(2, 26) = 2.09, p >.05. A one-way ANOVA omparing loomotion during the downshift, however, was signifiant, F(2, 26) = 22.49, p <.001. Planned omparisons showed that both 32-4 1-32 and 32-4 4-32 engaged in more nose-down loomotion than the ontrol group during the downshift, F(l, 26) = 29.90,/> <.001, andf(l, 26) = 37.17,p <.001. To show that it is the downshift itself, and not the availability of alternate solutions that ause the inrease in loomotion, we onduted a planned omparison between shifted groups, showing no differenes between them, F(l, 26) < 1.0. Finally, a one-way ANOVA omparing loomotion during the upshift showed no differenes between groups, F(2, 26) = 2.34. As in the previous experiment, nose-down loomotion was evoked by the downshift, but not the upshift. In addition, the loomotor response did not depend on the availability of alternate solutions, but ourred simply as the onsequene of the downshift. Loation Changes To show that the loomotor response resulted in signifiant traversal of the maze, we ompared groups on the number of loation hanges made during the preshift, the downshift, and the upshift. Figure 5C shows 2-day averages of hanges in the number of loation hanges during the preshift, the downshift, and the upshift for eah group. A one-way ANOVA of preshift data showed a trend for greater loation hanges by the ontrol group, whih fell just short of statistial signifiane, F(2, 26) = 3.33, p >.05. In ontrast, a one-way ANOVA omparing loation hanges during the downshift was signifiant, F(2,26) = 17.61,p <.001. Planned omparisons showed that both 32-4'-32 and 32-4 4-32 made more loation hanges than the ontrol during the downshift, F(l, 26) = 31.69, p <.001, and F(l, 26) = 19.19, p <.001, respetively, but did not differ amongst themselves, F(l, 26) = 1.86, p >.05. Thus, shifted animals traversed the maze more than ontrols did during the downshift only, and the number of these shift-indued traversals did not depend on the availability of alternate solutions.
4Hl SEARCH AND INCENTIVE DOWNSHIFTS 163 <u CB to d> O> C <B O O <B O O Figure 5. (A) Mean intake of surose solutions (in ml) (±SEM) aross suessive days for shifted and unshifted groups (Experiment 3). (B) Mean seonds engaged in nose-down loomotion (±5 M) for eah group during the preshift, the downshift, and the upshift in 2-day bloks (Experiment 3). (C) Mean number of entries to new loations (±5 M) for eah group during the preshift, downshift, and upshift in 2-day bloks (Experiment 3). (D) Mean number of bouts of general orienting responses (i.e., nose-up posture and open rearing) for eah group (±5 M) during the preshift, downshift, and upshift in 2-day bloks (Experiment 3). 32-4'-32 = downshifted in one loation; 32-4 4-32 = downshifted in four loations; 4-4M = unshifted in one loation. Orienting Responses To show that only downshifts indued orienting responses along with loomotor responses and maze traversals, we ompared bouts of orienting responses between groups during the preshift, the downshift, and the upshift. Figure 5D shows 2-day averages of hanges in bouts of open rearing and nose-up during the preshift, the downshift, and the upshift for eah group. A one-way ANOVA omparing bouts of orienting during the preshift showed no differenes between groups, F(2, 26) < 1.0. A one-way ANOVA, omparing bouts of orienting during the downshift, however, showed signifiant group differenes, F(2, 26) = 16.71, p <.001. Again, planned omparisons showed that both 32-4 ] -32 and 32-4 4-32 made more orienting responses than did the ontrol group, F(l, 26) = 31.19, p <.001, and F(l, 26) = 16.66, p <.001, respetively, whereas no differenes existed between shifted groups, F(l, 26) = 2.26, p >.05. Beaker Sampling Finally, to show that inreases in loomotion, traversals, and orienting were direted, and not simply energized behaviors, the sampling of old and new loations was ompared between groups during the downshift and the upshift. Figure 6 shows sampling behavior during the downshift and the upshift. The left panel shows sampling
164 PECORARO, TIMBERLAKE, AND TINSLEY TP CW1 CW2 CW3 Downshift TP 32-41-32 0 32-4 4-32 4-4 4-4 CW1 CW2 CW3 Upshift Figure 6. Mean number of sampling bouts (±SEM) during the first day of the downshift (left panel), and the mean number of sampling bouts per platform during the first day of the upshift (right panel; Experiment 3). 32-4 ] -32 = downshifted in one loation; 32-4 4-32 = downshifted in four loations; 4-4 4-4 = unshifted in one loation; TP = target platform; CW = lokwise. behavior on eah platform for eah group during the first day of downshift. The right panel shows sampling on eah platform during the first day of the upshift. In spite of having differing numbers of beakers to sample, no differenes in sampling frequeny were found between Groups 32-4'-32 and 32-4 4-32, and their data were pooled. A 2 X 2 (Group X Day) ANOVA on shifted and unshifted groups showed main effets of group, F(l, 28) = 18.53, p <.001, and day, F(l, 18) = 103.42, p <.001, and a Group X Day interation, F(l, 28) = 44.88, p <.001. Planned omparisons showed a greater frequeny of sampling by the shifted groups than the unshifted group during the downshift, F(l, 28) = 29.42, p <.001, whereas during the upshift, the shifted groups displayed less sampling than the unshifted group, F(l, 28) = 7.5S,p <.02. With respet to the spatial distribution of sampling by the shifted groups during the downshift, a one-way ANOVA showed a main effet of platform, F(3, 54) = 20.81, p <.001. A planned omparison of the target versus nontarget platforms showed that more sampling ourred on the target platform than on the average nontarget platform, F(l, 18) = 35.64, p <.001. One again, the target platform was not relatively aversive during the downshift, but was rather more attrative. Disussion The main finding of Experiment 3 was that the mobilization of searh behaviors following downshifts in Experiments 1 and 2 did not depend on the availability of alternative inentives. In fat, the rats in Experiment 3, having no alternatives other than the downshifted solution in the target arm, showed levels of loomotion, orientation, and sampling indistinguishable from that of rats having several other solutions available. The similarity in searh responses to inentive downshifts by groups having no alternative solutions and groups having three alternatives supports the hypothesis that the repertoire of searh responses is largely evoked by the inentive downshift itself, rather than being eliited or strengthened by the presene of the alternative solutions. The seond finding of Experiment 3 was that the large sale and sudden hange in searh behavior mounted in response to downshifts in reward is equally rapidly abated by reintrodution of the original solution. This observation is onsistent with the idea that the rats are searhing for a partiular ommodity, even though the ability of other ommodities similar in preferene, alories, or taste to end searh behaviors was not tested. At minimum, the reintrodution of the previous ommodity was suffiient to end searh. General Disussion We have previously observed rats in antiipatory ontrast studies exhibiting searh behaviors on shift days even in small home ages, lifting beaker holders and looking behind them when they disovered a poor solution. The urrent experiments on a four-arm radial maze showed a more omplete repertoire of searh behaviors following downshifts, inluding nose-down loomotion, loation hanges, open rearing, nose-up postures, and inreased sampling of other beakers and areas of the apparatus. Clearly, the derease in ingestion did not simply permit shifted animals to engage in random operants: Shifted rats systematially explored all spaes in the maze and sampled all ontainers. The behavior systems view aounts for the inverse relation between intake and searh in ontrast studies by assuming that a feeding system orhestrates both onsummatory and appetitive behaviors during hanges in food availability through ompetition between motivational modes within the system (in addition to any peripheral response ompetition that may happen to our). Whereas regular or improved food availability entrains or maintains onsummatory modes supporting ingestive responses, downshifts in food step the system bakward into foal or general searh modes, supporting a number of behaviors that diret the animal toward finding new food soures. The onsistent pattern of responses shown aross experiments suggests that intake suppression is a onsequene of a shift in motivational modes following food redution. Elliott's (1928) results, the majority of Flaherty's results (Flaherty et al., 1978, 1979; Flaherty, 1991; Flaherty, Tronoso, & Deshu, 1979), and the present results all support this onlusion. The present experiments also showed that the arm and platform formerly ontaining 32% surose did not appear to beome aversive or inhibit behavior. Former target areas still remained more attrative than did other areas of the maze. In all three experiments, rats preferred returning to and sampling from the areas in whih they had the most prior experiene with the rih surose solution. This plae preferene resulted from a pattern of behavior resembling arearestrited searh disussed in the ethologial literature (e.g., Bell, 1991). These findings are pertinent to interpretations of ontrast (e.g., Amsel, 1958; 1992; Blak, 1968; Daly, 1974b) that assume that the downshift is aversive, or that ontrast is mediated by the aversiveness of the downshift. The failure of loation avoidane to develop and the systemati nature
SEARCH AND INCENTIVE DOWNSHIFTS 165 of both searh and sampling suggests that esape behaviors by rats in ontrast experiments may atually be searh behaviors evoked by the absene of the expeted reward. From a behavior systems perspetive, we would argue that the redued intake and inreased systemati exploration, espeially in loations where plae preferenes had developed, represented a shift in motivation from onsummatory to foal and general searh modes, eah supporting different lasses of behaviors. It remains to be seen whether former target areas on the maze eventually would have beome relatively inhibitory if further training with the lesspreferred solution had been given during the downshift. At the very least, a frustration aount of ontrast does not predit the evoation of a repertoire of searh behaviors and the ontinued preferene for original target loations following inentive shifts during onsummatory suppression. An expliit attempt to ondition frustration to suessful loomotor searh in Experiment 1 by onstantly varying the loation of the target arm failed to inrease the ativity levels or the persistene of searh following the downshift. From a frustration view, failure to find food after running down the arms of the maze should produe frustration, whih should beome onditioned to eventually finding food on the maze, so when the downshift ours, one would expet an exitatory effet of frustration on instrumental responding. This sort of effet was not evident in searh behaviors within the first 2 days of the downshift. The group trained in a fixed loation showed as many loation hanges as the group trained in varying loations. The fat the 32 VL-4 showed slightly more nose-down loomotion is offset by the fat that this group showed similar effets before the shift and that 32 FL-4 showed more sampling, following the downshift. The additional sampling by 32 FL-4 resulted primarily from additional entries to the former target arm. Thus, the use of randomly varying surose loations provided no strong evidene for the development of instrumental persistene beause of daily frustration. A straightforward interpretation of this data is that 32 FL-4 was trained to prefer a partiular platform and thus inreased its sampling there, whereas 32 VL-4 was trained to loomote for food, resulting in slightly greater loomotor responses during the shift: Shifted groups merely superimposed previously pratied patterns on their evoked searh, whih was robust in both ases, despite these relatively minor effets of training. We would argue that fixed-loation training produed greater onditioning of foal searh modes to the target platform than did variable training, and was expressed as area-restrited searh and a plae preferene, whereas the variable-loation training resulted in greater onditioning of a general searh mode, whih was expressed as greater loomotor ativity. The results also showed that alternate soures of food were not required for exhaustive searh of the maze. The possibility remained after Experiments 1 and 2 that the searh behavior found after downshifts resulted only from the introdution of novel soures of food (though one would also expet the 4% ontrol group to show similar inreases in searh when novel solutions were introdued). By diretly omparing groups downshifted to single or multiple 4% solutions, it was shown that the additional solutions had no failitating effets on searh. Searh was equally robust regardless of the availability of alternative solutions. The group downshifted in a single loation showed at least as muh loomotion, orienting, entries, and sampling as the group shifted in several loations. The thoroughness of sampling for both shifted groups indiates the systemati nature of the searh, and it suggests the onditioning of foal searh modes to loal food ues, suh as platforms and beakers. A general finding in Experiments 2 and 3 was that loomotion, orienting, sampling, and intake suppression were not evoked by upshifts in reward. Experiment 3 showed that one searh behaviors were evoked by downshifts, a subsequent upshift urtailed this searh immediately. That upshifts in onentration produe none of these measures of ontrast supports Flaherty's (1991) assertion that neither generalization or neophobia an aount for NCEs. Furthermore, that all of these measures of ontrast eased immediately on the reintrodution of the 32% surose shows that the behaviors were related speifially to dereases in solution onentration. Capaldi (1972) showed that rats given massed trials (during whih primary frustration should not dissipate) showed a greater NCE than rats given distributed trials (during whih primary frustration should dissipate, leaving only onditioned frustration). He ontended that onditioned frustration was of less importane to NCE than primary frustration, even though frustration was never diretly measured. Another way of looking at his data, however, is to assume that the food given to rats in massed trials represented a larger and hene more important part of their daily energy budget than that of rats given distributed trials. Therefore, the downshift represented a greater threat to their energy budget and alled forth a greater searh response, whih in turn may have interfered more with riterion responses suh as runway speed. The general fat that inreasing experiene with rih foods leads to stronger ontrast is well known and supports an inentive averaging onept of ontrast (Flaherty, Beker, & Osborne, 1983; Peters & MHose, 1975). The inentive averaging onept is onsistent with foraging theories that assume that animals base foraging deisions, suh as when to abandon a path, on omparisons between urrent intake and averages of intake based on past experiene (e.g.,.chamov, 1976). The interpretation that NCEs largely represent a shift hi motivational modes would predit that greater experiene with better foods should all forth a greater searh response following a downshift. Data from shifted groups having different histories with the 32% surose in Experiment 2 are not inompatible with this view, though other interpretations are possible. This hypothesis requires further study. The urrent findings fit well into both Flaherty's (1996) multistage hypothesis of ontrast, whih suggests that animals do attempt to reover the missing ommodity following a downshift, and a behavior systems view, whih argues that intake suppression and searh are related expressions of the same goal-direted behavior orhestrated by the
166 PECORARO, TIMBERLAKE, AND TINSLEY feeding system. The main differenes between Flaherty's multistage model and a behavior systems approah are as follows: (a) the multistage hypothesis is sequential stage model speifi to suessive negative ontrast, whereas the behavior systems approah was developed as a more general model of animal learning and behavior in the laboratory; (b) the behavior systems approah speifies the relations between motivational systems and topographies of behaviors seen in ontrast studies; () behavior systems approah has not been formulated with respet to emotions or the reovery proess following sudden food redutions, whereas the multistage view suggests a reovery role for emotion-like physiologial states and suggests plausible mehanisms by whih this reovery proess may take plae. The behavior systems view and the multistage aount are not inompatible. In order to integrate adaptive and emotional aounts of NCEs, one might onsider, "What is the reprodutive advantage of a rare, frustrated mutant?" The emotional responses and the reovery sequene outlined by Flaherty (1996) may play an integrative role in adaptive responses to hanges in feeding opportunities, espeially insofar as the limbi system is known to play dominant roles in emotion, motivation, energy regulation, and memory. For example, adrenal hormones that are evoked by food redution and that are widely known to be involved in energy regulation, also may be involved in modulating memories for shifts in the amount or loations of foods (e.g., see Cahill, Roozendaal, & MGaugh, 1997) at a time when new memories may be ritial to survival. A systems view is potentially well suited to make suh preditions, insofar as inferenes regarding behavioral funtions suggest adaptive ontrol strutures. Although it is not being laimed that searh behaviors ause intake suppression, the relations between ingestion and searh behaviors in this and other studies suggest that the ommon motivation of feeding regulation underlies them all. Other NCEs, suh as esape responses, hanges in runway speeds, hoie errors, and even hanges in adrenal hormones, may be usefully interpreted as part of a repertoire or system of responses related to food gathering or energy regulation. Critial variables influening NCEs, suh as the degree of disparity between rewards, deprivation levels, and rates of energy intake, might also be usefully examined in terms of feeding-system issues. In short, a ase an be made for reevaluating disparate responses to inentive downshifts in terms of response systems, thereby potentially providing a more oherent aount of NCEs that inludes onsummatory suppression, searh repertoires, hormonal responses, memory formation, and emotions. Referenes Adelman, H. M., & Maatsh, J. L. (1956). Learning and extintion based upon frustration, food reward, and exploratory tendeny. Journal of Experimental Psyhology, 52, 311-315. Amsel, A. (1951). A three-fator theory of inhibition: An addition to Hull's two-fator theory. Paper presented at Southern Soiety for Philosophy and Psyhology Meetings, Roanoke, VA. Amsel, A. (1958). The role of frustrative nonreward in nonontinuous reward situations. Psyhologial Bulletin, 55, 102-119. Amsel, A. (1962). Frustrative nonreward in partial reinforement and disrimination learning: Some reent history and a theoretial extension. Psyhologial Review, 69, 306-328. Amsel, A. (1992). Frustration theory: An analysis of dispositional learning and memory. Cambridge, England: Cambridge University Press. Bell, W. J. (1991). Searhing behavior: The behavioral eology of finding resoures. London: Chapman & Hall. Benefield, R., Osos, A., & Ehrenfreund, D. (1974). Role of frustration in suessive positive ontrast. Journal of Comparative and Physiologial Psyhology, 86, 648-651. Blak, R. W. (1968). Shifts in magnitude of reward and ontrast effets in instrumental and seletive learning: A reinterpretation. Psyhologial Review, 75, 114-126. Cahill, L., Roozendaal, B., & MGaugh, J. L. (1997). The neurobiology of memory for aversive emotional events. In M. E. Bouton, M. S. Fanselow, et al. (Eds.), Learning, motivation, and ognition: The funtional behaviorism of Robert C. Holies (pp. 369-384). Washington, DC: Amerian Psyhologial Assoiation. Capaldi, E. J. (1972). Suessive negative ontrast: Intertrial interval, type of shift, and four soures of generalization derement. Journal of Experimental Psyhology, 96, 433^38. Charnov, E. L. (1976). Optimal foraging: The marginal value theorem. Theoretial Population Biology, 9, 129-136. Coover, G. D. (1983). Positive and negative expetanies: The rat's reward environment and pituitary adrenal ativity. In H. Ursin & R. Murison (Eds.), Biologial and psyhologial basis of psyhosomati disease: Advanes in the biosienes (Vol. 42, pp. 45-60). Oxford, England: Pergamon Press. Crespi, L. P. (1942). Quantitative variation in inentive and performane in the white rat. Amerian Journal of Psyhology, 55, 467-517. Daly, H. B. (1974a). Arousal of frustration following gradual redutions in reward magnitude, Journal of Comparative & Physiologial Psyhology, 86, 1149-1155. Daly, H. B. (1974b). Reinforing properties of esape from frustration. In G. H. Bower (Ed.), The psyhology of learning and motivation (pp. 187-232). New York: Aademi Press. Ehrenfreund, D., & Badia, P. (1962). Response strength as a funtion of drive level and pre- and postshift inentive magnitude. Journal of Experimental Psyhology, 63, 468-471. Elliott, M. H. (1928). The effet of hange of reward on the maze performane of rats. University of California Publiations in Psyhology, 4, 19-30. Flaherty, C. F. (1982). Inentive ontrast: A review of behavioral hanges following shifts in reward. Animal Learning & Behavior, 10, 409-440. Flaherty, C. F. (1991). Inentive ontrast and seleted animal models of anxiety. In C. F. Flaherty & L. Dahowski (Eds.), Current topis in animal learning: Brain, emotion, and ognition (pp. 207^137). Hillsdale, NJ: Erlbaum. Flaherty, C. F. (1996). Inentive relativity. Cambridge, England: Cambridge University Press. Flaherty, C. F, Beker, H., & Osborne, M. (1983). Negative ontrast following regularly inreasing onentrations of surose solutions: Rising expetations or inentive averaging? The Psyhologial Reord, 33, 415-420. Flaherty, C. F, Beker, H. C., & Pohoreky, L. (1985). Correlation of ortiosterone elevation and negative ontrast varies as a funtion of postshift day. Animal Learning & Behavior, 13, 309-314. Flaherty, C. F, Blitzer, R., & Collier, G. H. (1978). Open field behaviors eliited by reward redution. Amerian Journal of Psyhology, 91, 429-443. Flaherty, C. F, Powell, G., & Hamilton, L. W. (1979). Septal lesion, sex, and inentive shift effets on open field behavior of rats. Physiology and Behavior, 22, 903-909.
SEARCH AND INCENTIVE DOWNSHIFTS 167 Flaherty, C. P., Tronoso, B., & Deshu, N. (1979). Open field behaviors orrelated with reward availability and reward shift in three rat strains. Amerian Journal of Psyhology, 92, 385-400. Hearst, E., Bottjer, S. W., & Walker, E. (1980). Conditioned approah-withdrawal behavior and some signal-food relations in pigeons: Performane and positive vs. negative assoiative strength. Bulletin of the Psyhonomi Soiety, 16, 183-186. Lombardi, B. R., & Flaherty, C. F. (1978). Apparent disinhibition of suessive but not of simultaneous ontrast. Animal Learning & Behavior, 6, 30=42. MHose, J. H., & Peters, D. P. (1975). Partial reward, the negative ontrast effet, and inentive averaging. Animal Learning & Behavior, 3, 239-244. Meler, T., & Timberlake, W. D. (1985). Poison avoidane and path (loation) seletion in rats. Animal Learning & Behavior, 13, 60-68. Miller, W. L., & Tyrrell, J. B. (1995). The Adrenal Cortex. In P. Felig, J. D. Baxter, & L. A. Frohman (Eds.), Endorinology and Metabolism (3rd ed., pp. 555-712). New York: MGraw-Hill. Owen, O. E. (1989). Starvation. In L. J. DeGroot, (Ed.), Endorinology, Vol. 3 (2nd ed., pp. 2282-2293). Philadelphia, PA: W. B. Saunders. Peters, D. P., & MHose, J. H. (1975). Effets of varied preshift reward magnitude on suessive negative ontrast effets in rats. Journal of Comparative and Physiologial Psyhology, 86, 85-95. Shanab, M. E., & Ferrell, H. J. (1970). Positive ontrast in the Lashley maze under different drive onditions. Psyhonomi Siene, 20, 31-32. Silva, K. M., & Timberlake, W. D. (1997). Abehavior systems view of responding to probe stimuli during an interfood lok. Animal Learning & Behavior, 26, 313-325. Silva, F. J., Timberlake, W. D., & Gont, R. S. (1998). Spatiotemporal harateristis of serial CSs and their relation to searh modes and response form. Animal Learning & Behavior, 26, 299-312. Spear, N. E., & Spitzner, J.H. (1966). Simultaneous and suessive ontrast effets of reward magnitude in seletive learning. Psyhologial Monographs, 80 (10, Whole No. 618). Spear, N. E., & Spitzner, J. H. (1969). Influene of degree of training and prior reinforer magnitude on ontrast effets and resistane to extintion within S. Journal of Comparative & Physiologial Psyhology, 68, 427-433. Spene, K. W. (1956). Behavior theory and onditioning. New Haven, CT: Yale University Press. Timberlake, W. D., & Luas, G. A. (1989). Behavior systems: An integrative approah. In S. B. Klein & R. R. Mowrer (Eds.), Contemporary theories of learning: Instrumental onditioning theory and the impat of biologial onstraints on learning (pp. 237-276). Hillsdale, NJ: Erlbaum. Tinkelpaugh, O. L. (1928). An experimental study of representative fators in monkeys. Journal of Comparative Psyhology, 8, 197-236. Reeived February 23, 1998 Revision reeived Otober 12,1998 Aepted Otober 13, 1998 New Editors Appointed, 2000-2005 The Publiations and Communiations Board of the Amerian Psyhologial Assoiation announes the appointment of three new editors for 6-year terms beginning in 2000. As of January 1,1999, manusripts should be direted as follows: For Experimental and Clinial Psyhopharmaology, submit manusripts to Warren K. Bikel, PhD, Department of Psyhiatry, University of Vermont, 38 Flether Plae, Burlington, VT 05401-1419. For the Journal of Counseling Psyhology, submit manusripts to Jo-Ida C. Hansen, PhD, Department of Psyhology, University of Minnesota, 75 East River Road, Minneapolis, MN 55455-0344. For the Journal of Experimental Psyhology: Human Pereption and Performane, submit manusripts to David A. Rosenbaum, PhD, Department of Psyhology, Pennsylvania State University, 642 Moore Building, University Park, PA 16802-3104. Manusript submission patterns make the preise date of ompletion of the 1999 volumes unertain. Current editors, Charles R. Shuster, PhD; Clara E. Hill, PhD; and Thomas H. Carr, PhD, respetively, will reeive and onsider manusripts through Deember 31, 1998. Should 1999 volumes be ompleted before that date, manusripts will be redireted to the new editors for onsideration in 2000 volumes.