526011/955/ W gflIMfllfi/IWEWMMFW ”my Michigan State J. University This is to certify that the dissertation entitled Selective Interference on Conditioned Odor and Taste Aversions: A Test of the Indexing Hypothesis of Odor Potentiation presented by Silvia von Kluge has been accepted towards fulfillment of the requirements for Ph. D. (,3ng Psychology W’ ’ ’~ , Ema/gym ’ MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove thIo checkout from your record. TO AVOID FINES return on or baton date due DATE DUE DATE DUE DATE DUE J L_Ji_ MSU Is An Afllnndlvo Action/Equal Oppodunlty Institution ammo-9.1 SELECTIVE INTERFERENCE ON CONDITIONED ODOR AND TASTE AVERSIONS: A TEST OF THE INDEXING HYPOTHESIS OF ODOR POTENTIATION by Silvia von Kluge A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements ' for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1990 m05§04 ABSTRACT SELECTIVE INTERFERENCE ON CONDITIONED ODOR AND TASTE AVERSIONS: A TEST OF THE INDEXING HYPOTHESIS OF ODOR POTENTIA’I'ION by Silvia von Kluge A series of experiments were run in order to determine whether odor takes on the associative properties of taste in food avoidance learning. The theoretical position that was tested is the "indexing hypothesis" put forward by Rusiniak and Garcia (1979). Their hypothesis was formulated to explain the "taste-potentiation" effect in which a taste stimulus facilitates the strength of an odor aversion when odor and taste are presented as a compound before lithium chloride induced illness. Three experiments used an interference design, in which either external stimulation (exteroceptive manipulations) or internal stimulation (interoceptive manipulations) was presented during the delay between consumption of the taste-odor target compound and lithium chloride induced illness, to see if they would disrupt conditioning to the taste and / or odor component. According to the indexing hypothesis, which suggests that odors take on the qualities of taste by taste-potentiation, disruption should occur only when internal events are presented in this interval since it has been shown that taste is not disrupted by external events. Experiment 1 found disruption of the odor only by exteroceptive interference manipulations. Experiment 2, in which weaker conditioning parameters were used, found interference on both odor and taste by both types of interference manipulations. It was also noted that there was intense behavioral arousal in both Experiment 1 and Experiment 2, probably as a result of the experimental schedule. Therefore, in Experiment 3 both weak and strong conditioning parameters were used in a procedure that reduced the arousal of the subjects. Under these conditions only interoceptive interference manipulations disrupted conditioning to the odor component of the target compound. This was true under both weak and strong conditioning parameters. It was suggested that the class of events that interferes the conditioning of a taste-potentiated odor depends on the arousal level of the organism. When subjects were highly aroused (Experiments 1 & 2), disruption by external events was evident. Activation of the sympathetic nervous system may have facilitated additional attention towards external events. However, when arousal was reduced, as in Experiment 3 disruption by internal events was more pronounced. Interactions between external threats and feeding may be commonplace. Given a behaviorally aroused circumstance, interaction between external defense and interoceptive processes are engaged and the odor cues tend not to be processed as taste cues, contrary to the indexing hypothesis. Given a behaviorally quiet situation, odor cues do operate like taste cues in accordance with the indexing hypothesis. ACKNOWLEDGEMENTS I would like to express my gratitude to those people whose help in this enterprise was invaluable: Dr. Ken Rusiniak, for your patience, wisdom, hard work, and endless hours of rat running and conversation. I am deeply grateful for the opportunity to have worked with you, and look forward to many more collaborative quests in search of the secrets of associative mechanisms. Thank you from the bottom of my heart. Dr. Tony Nunez, who makes it all seem so easy and so much fun. Thank you for your many helpful suggestions, your support, your enthusiasm, and your participation in this project. Dr. Mark Rilling, for sharing countless books, references, and ideas. The time I spent in your pigeon lab taught me a great deal, and provided a valuable perspective. Thank you for your support in this enterprise. Dr. Tom Carr, for support when I needed it, and challenge, even when I thought I didn't! Thanks for your confidence in me, our many lively conversations, and the inspiration you provided. Dr. Dave Irwin, thanks for the soft touch when it was most appreciated, and your quiet guidance and confidence. It was a joy to have you on my committee. Finally, I would like to thank the rats who served as subjects in my experiments. They are wonderful creatures, curious, patient, and always punctual. TABLE OF CONTENTS LIST OF TABLES AND FIGURES .................................................................................. vi William James: Attention . - - - - _ ................................. 1 Conditioned flavor aversions and general processes learning theory ................................................................................................................................... 2 Odor potentiation and memorial indexing ................................................................. 5 Interference, selective interference and potentiation ................................................ 10 General implications ........................................................................................................ 14 General experimental design .......................................................................................... 15 Experiment 1 ....................................................................................................................... 17 Procedure ............................................................................................................................. 22 Results ................................................................................................................................. 24 Discussion ........................................................................................................................... 29 Experiment 2. ...................................................................................................................... 32 Procedure ............................................................................................................................. 35 Results .................................................................................................................................. 38 Discussion ........................................................................................................................... 42 Interim summary .............................................................................................................. 46 Experiment 3 ....................................................................................................................... 47 Procedure ............................................................................................................................. 49 Habituation Trials ............................................................................................................. 50 Results ................................................................................................................................. 51 ii The target stimuli (01 Gr T1) ........................................................................................... 51 The interference stimuli (02 6: T2) ............................................................................... 58 Discussion ........................................................................................................................... 61 Comparison and contrast of Experiment 3 with Experiments 1 & 2. ..................... 61 General DISCUSSIOD - - ........... _ -- ....... 68 List of References ............................................................................................................... 73 iii LIST OF TABLES AND FIGURES Tables Table 1 .................................................................................................................................. 18 Table 2. ................................................................................................................................ 20 Table 3 ................................................................................................................................. 36 Figures Figure 1 ................................................................................................................................ 25 Figure 2. ............................................................................................................................... 28 Figure 3 ................................................................................................................................ 31 Figure 4 ................................................................................................................................ 39 Figure 5 ................................................................................................................................ 43 Figure 6 ................................................................................................................................ 45 Figure 7 ................................................................................................................................ 52 Figure 8 ................................................................................................................................ 54 Figure 9 ................................................................................................................................ 59 Figure 10 .............................................................................................................................. 60 Figure 1 1 .............................................................................................................................. 62 Figure 12. ............................................................................................................................. 63 Figure 13 .............................................................................................................................. 64 iv ATTENTION Millions of items of the outward order are present to my senses which never properly enter into my experience. Why? Because they have no interest for me. My experience is what I agree to attend to. Only those items which I notice shape my mind - without selective interest, experience is an utter chaos. Interest alone gives accent and emphasis, light and shade, background and foreground - intelligible perspective, in a word. It varies in every creature, but without it the consciousness of every creature would be a gray chaotic indiscriminateness, impossible for us even to conceive. Such an empiricist writer as Mr. Spencer, for example, regards the creature as absolutely passive clay, upon which 'experience' rains down. The clay will be impressed most deeply where the drops fall thickest, and so the final shape of the mind is molded. Give time enough, and all sentient things ought, at this rate, to end by assuming an identical metal constitution - for 'experience', the sole shaper, is a constant fact, and the order of its items must end by being exactly reflected by the passive mirror which we call the sentient organism. If such an account were true, a race of dogs bred for generations, say in the Vatican, with characters of visual shape, sculptured in marble, presented to their eyes, in every variety of form and combination, ought to discriminate before long the finest shades of these peculiar characters. In a word, they ought to become, if time were given, accomplished connoisseurs of consummation. Surely an eternity of experience of the statues would leave the dog as inartistic as he was at first, for the lack of an original interest to knit his discriminations on to. Meanwhile the odors at the bases of the pedestals would have organized themselves in the consciousness of this breed of dogs into a system of 'correspondences' to which the most hereditary caste of custodi would never approximate, merely because to them, as human beings, the dogs interest in those smells would for ever be an inscrutable mystery. [Note From The Principles of Psychology Vol. 1 (pp 402-403) by William James, 1890, New York: Henry Holt and Company. Emphasis his] SELECTIVE INTERFERENCE ON CONDITIONED ODOR AND TASTE AVERSIONS: A TEST OF THE INDEXING HYPOTHESIS OF ODOR POTENTIATION The phenomenon of taste aversion learning has received a great deal of attention in the past three decades. The initial impact of this now classic research focused on the ways taste-aversion learning violated traditional, i.e., general process, laws of learning. Three basic features of taste aversion learning attracted attention. These were: 1) One trial learning. Taste aversion conditioning could occur with a single taste-illness pairing (Garcia 8: Ervin, 1968; see Domjan, 1983; and Logue, 1979 for reviews), 2) Long delay learning. Taste aversions could be conditioned with delays of up to an hour or more between the presentation of the taste conditioned stimulus (CS) and the poison induced illness unconditioned stimulus (US); 3) Selective association. Taste was a much better one than an audio-visual cue when the consequence was illness and conversely, an audio-visual one was a more effective cue than taste when the consequence was electric shock. One-trial long delay learning and selective associations formed the cornerstone to one alternative to general process theory, the "biological constraints" position which gained credibility and empirical support over the years. But in addition to conditioned taste aversion, there was other evidence that did not support the "general process" view of learning. For example, one trial passive avoidance learning, in which a place or bottle spout was paired with shock, and long delay learning in the T-maze and runway (Capaldi, 1967; Lett, 1973, 1974; Petrinovich Gr Bolles, 1957) had been known for years. Further, the work of the Brelands (1960) with "instinctive drift", and Bolles (1970) with "species specific defense reactions" had shown that there were constraints on the response a species could learn to make in different situations and these biases depended on the ecological niche in which that species evolved. But it was the phenomenon of selective association that made the taste aversion research so powerful and problematic for general process accounts of learning and memory, and served as the cornerstone of the biological constraints view. Stimuli that were very effective signals for avoidance in a species for one type of consequence (e.g. tastes as cues for illness, audio-visual cues as signals for shock) were less than efficient as signals for a different type of consequence (taste as a signal for shock was difficult to establish, as were audio-visual stimuli as cues for illness). Thus it could not be argued that the cues differed in salience, that the organism did not attend to them, or that one US was stronger than another US. The symmetry of the experimental design forced the conclusion that it was the consequence, or type of US, that determined which stimulus the animals would select as the CS, and that stimuli that were readily associable in one set of circumstances were not necessarily equally associable in another circumstance. Garcia and Koelling (1966) and Garcia and Ervin (1968) suggested that selective association reflects a biological segregation of two different adaptive learning systems, one for the milieu interior and the other for the milieu exterior. The exteroceptive-cutaneous, or skin defense system is thought to respond to "exteroceptive" insult (as in an attack by a predator), involves cognitive and motor coping strategies (i.e. escape), and is biased towards association between exteroceptive C93 and somatic US's. The gustatory-visceral, or gut defense, system is thought to protect organisms from poisoning by toxic plants and prey, and responds to "interoceptive" insult (such as gastrointestinal upset). This system is biased toward gustatory stimuli and involves visceral feedback which Operates over long delays by altering the hedonic value of a food. Thus when an animal encounters a substance that has made it ill, negative affect keeps it from consuming that substance again. A clear example of the origins of a separation between visceral and somatic systems can be found in the neuroanatomy of the tiger salamander (Herrick, 1948). In Herrick's words, "This segregation of all sensory nerve fibers, except those of vision and olfaction, into only two receptive centers is the only well-defined localization of sensory functions present in the medulla oblongata. It corresponds with the fundamental difference in behavior between internal visceral activities and somatic sensori-motor activities that have external reference" (Herrick, 1956). That is, neural output from the skin defense system is thought to converge in dorsal brain mechanisms and engage the striated muscular movement involved in avoidance or attack. Visceral feedback from consumatory behavior converges in the ventral regions of the brain where hedonic alteration and digestive reflexes may occur as a consequences of ingestion. Vision and olfaction are considered "unbiased" or multiple function senses, and are thought to be available to either system. This view of a dichotomous structure is not limited to the tiger salamander, but is thought to be representative of all vertebrates (Garcia 8: Garcia y Robertson, 1985). Evidence for two separable functions in rats was provided by McGowan, Hankins, and Garcia (1972). They demonstrated that lesions of the lateral septum or the ventral hippocampus disrupted noise-shock learning but had no such effect on flavor aversion learning, and in fact enhanced acquisition of flavor aversions. Other limbic lesions disrupted both or neither type of learning. Similarly, very discrete lesions of primary taste cortex disrupted taste aversion learning more than noise-shock learning (Hankins, Garcia 6: Rusiniak, 1974). McGowan et al. suggest that lateral septal and ventral hippocampal lesions may remove the inhibitory influences of the exteroceptive system on the more basic internal associative mechanisms. Thus, the cue to consequence specificity observed in conditioning by shock or poison is thought to be a result of differential processing by the two defense systems. Mainstream learning theory was modified to incorporate several of the findings from taste aversion research. Temporal parameters were adjusted to include long delay learning, and it was suggested that the salience of cues interacts with reinforcement to produce one trial learning (Bolles, 1979; Dickinson & Mackintosh, 1978; Logue,1979; Rescorla, 1988; Seligman,1970). Researchers (and journals) were subsequently identified with one or the other point of view. ”General process" theorists were those in search of associative mechanisms that are consistent across response systems. "Biological constraints" theorists were those who advocated a more ethological approach, insisted that structure, function and neurological mechanisms were legitimate concerns for psychology, and argued that learning theory must be couched in an evolutionary adaptive framework in which biological structure and function are relevant. However, understanding the mechanism responsible for "one to consequence” learning was of importance to both groups. The difference between these points of view has faded somewhat during the last decade as learning theory has adjusted to challenges from within (ethology) and without (cognitive psychology), and researchers of both orientations continue to pursue taste aversion research. Odor pgtentiation and memorial indexing More recent data in the taste aversion literature challenged general process theories of compound conditioning. Prominent theories of the day posit that when a compound conditioned stimulus is paired with a single US, the elements that comprise the CS compete for associative strength with the US (Kamin, 1969; Mackintosh, 1974, 1975; Pavlov, 1927; Rescorla 8: Wagner, 1972). These theories predict that the associatively stronger component of the compound disrupts, blocks, or overshadows, conditioning to the weaker component. When a taste or an odor is conditioned to be aversive by pairing with delayed LiCl, a substance which induces illness, the avoidance of taste is much stronger than the avoidance of an odor. That is, a taste cue presented alone is a more salient cue for illness than is an odor alone (Rusiniak, Hankins, Garcia, & Brett,1979). However, this study also demonstrated that when a compound taste-odor CS was presented prior to delayed LiCl poisoning, conditioning to the odor was stronger than if the odor had been conditioned alone, and that the resultant odor aversion was sometimes stronger than the aversion to the taste with which the odor had been compounded. This facilitation of odor conditioning was called potentiation and these data were directly opposite expectations based on prevailing learning theory. These results were widely replicated (Bouton, Jones, McPhillips, & Swartzentruber, 1986; Durlach 6t Rescorla, 1980; Lett, 1984; Palmerino Rusiniak, & Garcia, 1980; Rescorla & Curmingham, 1978; Rescorla & Durlach, 1981; Rusiniak, Palmerino, Rice, Forthman, dz Garcia, 1982), although failures to observe potentiation were also reported (Bouton 8: Whiting,1982; Mikulka, Pitts, & Philput, 1982). One possible explanation for this effect was that flavor stimuli, and perhaps other feeding cues, were just different from other stimuli, and would potentiate each other under any circumstance. Therefore the next issue became whether this relationship holds in a situation in which odor is a more salient cue than taste. It had been shown that odor was a more effective cue than taste for shock when conditioned alone (Hankins, Rusiniak, 8: Garcia,1976). However, when a taste-odor CS was presented in compound as a cue for shock, the associatively strong odor neither potentiated nor overshadowed the conditioning of the associatively weaker taste (Rusiniak et al.,1982). That is to say, when a taste-odor CS compound is presented, taste potentiates odor when the consequence is illness, but odor does not potentiate tastewhentheconsequenceisshock. Thisselectivenatureofthe potentiation phenomenon is reminiscent of cue to consequence limitations in taste aversion learning. Is potentiation another example of selective association, limited to odors and tastes and particular to the feeding system, or is potentiation an example of a general process that has analogues in other conditioning paradigms and can be accounted for by Pavlovian principles? Researchers in the general processes tradition suggest that potentiation is not a special phenomenon within the feeding system, but a variant of second order conditioning or perceptual learning of stimulus properties of an object (Rescorla & Cunningham, 1978; Rescorla & Durlach, 1981). It is their contention that within—compound associations between the taste and odor stimuli facilitate avoidance as follows. Since odor and taste were paired with each other in the taste-odor compound, they are thought to form an intracompound association. A separate association connects taste directly with illness; this occurs because presentation of the odor stimulus activates a mental representation of the taste, Thus when the odor is encountered, it activates a recollection of the taste, which is now aversive due to its association with illness. Since the odor-elicited taste representation is now active, consumption is inhibited in the presence of the odor. This point of view has had success in accounting for some properties of potentiation but failures when dealing with others (Rescorla & Cunningham, 1978; Rescorla & Durlach, 1981). Garcia and Rusiniak (1980) proposed a "memorial indexing hypothesis" which treats the potentiation effect as a variation of selective association; it is a specialized mechanism that "bridges" the internal and external conditioning domains. This hypothesis uses the dual system view as its starting point. Specifically, odor and taste play non-redundant and different biological roles in the feeding sequence. While odor may be high on a hierarchy of cues utilized in feeding it has primarily an exteroceptive fimction. It is used to locate and approach a foodstuff; it is not structurally or functionally connected directly to medullary swallowing, vomiting or digestive reflexes and hence by itself is a weak cue for poison. Further, odor serves many other external functions and may require "indexing" by the presenceoftastetoberecalledasafoodcueratherthanasanodorattributed to a place or mate. In contrast taste is neurally wired to visceral mechanisms in the medulla and plays primary an interoceptive frmction in adjusting palatability. According to the indexing hypothesis, an odor presented temporally and spatially contiguous to a taste is stored in memory as a food cue. An odor so indexed then gains access to the same visceral mechanisms as taste, so it may be protected from interference by other odor cues by virtue of its storage in the memory system of internal processes. Once such an association has been made the odor becomes an effective distal one since it allows the animal to avoid noxious foods without requiring that they be tasted before being rejected; thus odor becomes a functional food cue. A variety of behavioral experiments support the indexing hypothesis and suggest that potentiated odors are processed similarly to taste. Potentiated odor aversions are known to be resistant to extinction (Rusiniak et al., 1983), can be conditioned with the same long delay between presentation of the CS and illness (Rusiniak et al., 1982), increase as the intensity of the CS increases (Holder & Garcia, 1987b), and promote avoidance by instituting a change in the hedonic value of the CS (Garcia, 1989). These characteristics are not shared by other potentiated cues. That is, even though it has been shown that visual cues (Galef & Osborne, 1978), environmental cues (Best, Best, 8: Mickley, 1973) and auditory cues (Ellins & von Kluge, 1986) can be taste potentiated, they do not share the parameters with taste that odor does. Taste potentiation of visual cues fails with delays of longer than 15 min ( Galef & Osborne, 1978, Experiment II); environmental cues are difficult to condition, extinguish rapidly, and never suppress consumption to the extent that taste does (Ellins, Thompson, 8: Swanson, 1983); and auditory potentiation is difficult to demonstrate, it is weak, and extinguishes rapidly (Ellins & von Kluge, in press). Therefore it has been suggested that there exists a hierarchy of cue utilization that differs among species (Garcia, 1984; Bolles, 1979) and only odor has been shown to adopt the characteristics of taste. One hypothesis is that cues other than odors rarely gain access to the visceral system through the potentiation mechanism, and rather than inducing avoidance by way of a shift in palatability serve as more cognitive signals for avoidance (Ellins & von Kluge, 1986). There is also neurophysiological evidence for the segregation of gustatory-visceral and olfactory processes in mammals that supports the indexing hypothesis. Keifer, Rusiniak and Garcia (1982) found that lesions of the gustatory neocortex of rats disrupted taste aversions but did not block potentiated odor aversions. These results provided additional support for the position that the associative processes mediating taste and odor are neuroanatomically separate, and that while the gustatory neocortex may mediate some taste aversion processes it is not essential to the potentiation aspects of taste and odor conditioning. That this deficit is associative rather than sensory is supported by findings that these lesions do not disrupt detection thresholds or the preference-aversion function for basic taste stimuli (Kiefer, 1985). The limbic system has been one set of structures directly implicated in the potentiation process (Bermudez-Rattoni, Rusiniak & Garcia,1983; Bermudez-Rattoni, Sanchez, & Prado-Alcala, 1989). For example, Bermudez-Rattoni and his colleagues demonstrated that the amygdala plays a clear role in odor-illness learning and potentiation without much involvement in taste processes. Biochemical lesions of the amygdala or administration of cholinergic antagonists into the amygdala tended to disrupt 10 odor-illness and taste potentiated odor-illness conditioning without strong effects on taste-illness or odor-shock learning. Thus, there is evidence that there are neurologically distinct loci involved in the potentiation effect. Just as with the second-order and perceptual conditioning explanations for potentiation, the indexing view has had success in accounting for some aspects of potentiation and failure in explaining others (LoLordo & Droungas, 1989). More than a decade of research attempting to directly test the indexing hypothesis has failed to reveal the mechanisms responsible for potentiation. A great deal has been learned about the parameters that facilitate or attenuate the potentiation effect and how odor is incorporated into the regulation of feeding. The potentiation paradigm has been used with great success to understand species differences and the intricate rules that govern associative conditioning. However, a theory that accounts for all the data has not emerged. Therefore, the present project will take a different approach. Rather than looking directly at the mechanism of potentiation, we will use the phenomena of selective association to test the indexing hypothesis from another direction. With One of the assumptions of the indexing hypothesis is that in their basic operations the internal and external defense systems are not only anatomically separate but functionally independent (Garcia 6r Ervin, 1968). However, an odor conditioned in a compound with a distinctive taste will be "indexed" or coded as a food cue and be processed in the internal defense system. As a consequence its association with illness accrues properties similar to taste. Since the odor is being processed in the intemal defense system, it is then protected from interference by external events or other 11 odors thatoccurin thedelaybetweenodor andillnessbecausethisis afeature of the internal defense system. If, as suggested by Garcia and Rusiniak (1980), odor is "sequestered" into the internal associative system by taste, and if the internal defense system is relatively independent of the external defense system (Garcia 8r Erwin, 1968), then other exteroceptive events occurring during the delay between consumption of the taste-odor compound should not disrupt the odor-illness association. However, if other flavors or interoceptive events are encountered during the CS-US delay these events should produce some interference because they are also gustatory in nature. That is, interference effects should be "system dependent" with only interoceptive stimuli able to disrupt or interfere with odor-illness learning. Earlier research on the long delay nature of taste aversion learning have shown that there are interference effects between two complex flavors (Bond, 1983; Der-Karabetian & Gorry, 1974; Kalat &: Rozin,1971; 1972; Revusky,1971; Spear 8r Kucharski,1984). For example, Revusky (1971) and Kalat and Rozin (1971, 1972) conducted experiments that were similar in design to the present study. Revusky (Revusky, 1970; Revusky & Garcia, 1970) developed a concurrent interference theory of memory to explain all long delay learning. His model suggests that the strength of a reference association is an inverse function of the number of interference stimuli and their associative strength. That is, a CS-US association can tolerate a delay that is proportional to the number of relevant events that come between the CS and the US. Further, because target and interfering events can show reciprocal effects, a strong target stimulus protects itself through the weakening of the interference manipulation. In some ways his model is similar to indexing, but it does not explicitly deal with selective interference by different classes of stimuli. In the case where two novel tastes were presented in succession, it was found that the aversion to a flavor target (sugar water) was slightly attenuated when 12 followed by another taste (vinegar). Facilitation of the target taste could be achieved if pretraining had made vinegar (the interfering taste) a familiar safe cue, leaving the novel sweet water the most likely cause of illness. In this case thesweetwater was avoided moreintesting thanifthetastes hadbothbeen novel. Similar results were also reported by Kalat and Rozin (1971,1972) who demonstrated weak interference on the learning of target taste stimuli even when as many as three flavors were presented between the target flavor and illness. Taken together, these studies indicate that flavors introduced during the delay between a target flavor and the US in most cases attenuated, but did not eliminate, aversions to the target flavor, and that the greater the number of interpolated flavors the greater the attenuation. The experiments cited above were designed to answer general questions about the mechanisms that subserve long delay learning, not potentiation. Furthermore, the degree of independence between the internal and external defense systems has never been tested using an interference design. That is, no one has directly compared the effects of exteroceptive versus interoceptive cues presented during the CS-US interval, or tested whether there are selective interference effects of flavors on the odor or taste components of a complex flavor. Because the potentiation phenomena was as yet unknown, only complex flavors were the targets and only complex flavors were introduced during the CS-US delay. Research examining interference on taste-potentiated odors during the delay interval is scarce. It is known that odor potentiation is sensitive to spatial and temporal relationships between the taste and odor during the presentation of the stimulus compound (Coburn, Garcia, Kiefer, 8r Rusiniak, 1984; Holder & Garcia, 1987). Similarly, exposure to the odor and / or taste commnents of the flavor before and after training may have deleterious effects under some conditions (Holder, Leon, Yirmiya, 8r Garcia, 1987). While it has been shown that a taste cue will prolong the CS-US interval of a 13 concomittant odor to match that of the taste cue, we do not know whether the potentiation process is vulnerable to stimuli presented during the extended delay between compound CS presentation and illness, or whether interference with the conditioning of either component would depend on the type of interpolated experience. Further, there have been several failures to replicate the potentiation effect (Bouton & Whiting,1982; Milulka, Pitts, & Philput, 1982) that are troublesome to interpret because the reported procedures seem similar to those reported in successful replications. Conceivably, unreported procedural differences involving interference effects could be involved. It is possible that the odor and the taste components of an taste-odor compound would be affected differently by a gustatory or non-gustatory experience that occurred during the CS-US delay, or that the potentiation effect itself would be disrupted. Finally, potentiation of visual cues by taste in pigeons and rats does not extend the CS-US delay (Galef & Osborne, 1978; Westbrook Clarke, & Provost, 1980), so there is some reason to question the extent to which an indexing process would affect mechanisms responsible for tolerating long delays. One way to test these ideas would be to follow the presentation of a taste-odor compound by different classes of interpolated experiences, for example a flavor, an odor, an odor-flavor compound, or an important external event such as a shock or placement in a novel compartment. Since previous reports have shown that taste aversions are difficult to disrupt, it is likely that tastes would be relatively immune to interference, whether in compound with an odor or not. The effect on the odor component could follow a different pattern. If the indexing hypothesis is correct and odor is protected from interference by external events through its association with taste, then only the interoceptive manipulations should attenuate the target odor aversion. If both types of manipulations interfere with the target then we have evidence that the indexing hypothesis is not correct in all its 14 assumptions. Other possible but less interesting results are that the strongest aversions simply occur to the stimulus closest in time to the US, producing a general disruption of all aversions, or that neither exteroceptive nor interoceptive events interfere with conditioning of the target compound. According to the indexing hypothesis, the aversion for the odor component of a target compound will be as least as strong as the aversion to the taste that potentiated it. When exteroceptive stimuli occur during the CS—US interval, there should be little disruption of the taste component of the target, by virtue of its biological role, or the odor component, by virtue of indexing. In fact the crucial test of the indexing hypothesis is in evaluating the abith of the taste to provide protection for the odor from interference by exteroceptive events. However, when interoceptive stimuli occur, disruption should occur to the taste component of the target compound as well as to the odor component, and the pattern of disruption should be the same. That is, the indexing hypothesis predicts that only interoceptive interference ' manipulations would interfere with the association of the target and DC] poisoning, and that the odor component of the compound would be the most sensitive to such disruption. General implications These experiments are designed to shed light on the hypothesis that there are two separate but interacting learning systems, one gustatory-visceral and one exteroceptive-cutaneous, that tend to function independently, follow different temporal parameters, are adapted to select for particular sensory stimuli, and produce qualitatively different responses. Using an interference design in a potentiation paradigm we hope to determine whether the systems are functionally insulated, and what properties of the system are demonstrated by secondary or "indexed" cues. If the dual system theory is correct and the two systems operate by 15 different rules, then understanding the conditioning of affective responses will require a reevaluation of theories which draw on "laws of behavior" developed mainly through experiments designed to uncover universal mechanismsoflearning. If,ontheotherhand,therearetwosystems that function independently but are sensitive to one another, the conditions under which they interact and the nature of the relationship needs to be evaluated; see Rozin (1976) and Rozin & Schull (1989) for a discussion of multiple, or clusters of learning systems. Taste aversion and potentiation methodologies are ideally suited to addressing questions such as these (Garcia y Robertson & Garcia, 1985). Genral m’ ental design The purpose of these experiments was to examine the effect of a variety of interference treatments presented between exposure to a target taste-odor compound and the administration of LiCl 30 min later. Saccharin was chosen as the taste component of the target because it is a prepotent interoceptive cue easily associated with illness. Almond extract odor was chosen as the second component because it has an inherent exteroceptive role which is thought to change to an interoceptive function when presented together with saccharin taste (Rusiniak et al., 1979). This arrangement of stimuli and illness produces a reliable potentiation effect in our laboratory. Therefore we could compare interference effects on odor and taste under conditions known to produce potentiation. A range of treatments representing "interoceptive" and "exteroceptive" experiences were interpolated between the target taste-odor compound and illness as interfering events to see if intervening events would disrupt conditioning to the taste and / or odor targets. In order to enhance interference with the target, these interference manipulations were presented 15 min into the 30 min conditioning sessions to reduce temporal association with the target flavor and to maximize association with LiCl. 16 These parameters were chosen because taste potentiated odor aversions are quite robust at a 30 min. delay, and odor aversions are possible to condition at a delay of 15 min. As "interoceptive" interfering stimuli we chose flavors known to produce strong conditioning with an illness US and known to produce little or no stimulus generalization to the target stimuli. That is, interoceptive experiences were represented by discriminably different stimuli known to be readily associated with gastrointestinal upset. If these stimuli failed to interfere with the target aversion it could not be a result of their lack of salience in the conditioning situation. The "exteroceptive" interference manipulations were selected to simulate an important external event at least as complex as the target stimulus event. Places, or "contexts", are important stimuli for rats, and contextual change, or changing the place where a CS is presented, is known to disrupt conditioning (see Balsam & Tomie, 1985) as well as the retardation of taste aversion learning caused by the taste familiarity effect (Rudy, Rosenburg, & Sandell, 1977). Further, in a study by Krane and Robertson (1982) which was designed to distinguish between theories that could account for the familiarity effect, a change in context was used as a novel exteroceptive manipulation and was inserted between dual exposures to a single taste. They found this exteroceptive stimulation to be successful in disrupting learning in some conditions but not in others. According to the indexing hypothesis, odors are used by rats to identify a place. Therefore, an odor presented in a novel place would be expected to be indexed into the external defense system, and the exteroceptive manipulation would then be an external compound. Thus we used both a novel context alone and an odor in the novel context as exteroceptive cues. Finally, we used odor alone presented around the drinking spout. Such an arrangement does not fall neatly into either classification. An odor alone interference condition was included to allow comparison for the groups in 17 which a second odor was a part of a stimulus compound. It is referred to as an "unbiased" interference manipulation for the purposes of the present experiments; however, it was expected to function as a weak interoceptive cue since it was presented at the drinking source, a spatial relationship that facilitates odor illness learning, and there was no other explicit exteroceptive event to reinforce odor's external function. Five groups appear in each of the experiments that follow. Several additional interference manipulations and controls were run in Experiments 1 8r 2 and will be discussed in the section describing the desigi, rationale and results of each experiment. The five major conditions are listed in Table 1. They include one basic control, the target taste-odor compound followed by no interference manipulation (H20), to which all the other manipulations can be compared to determine the extent of the interference they produce. Two exteroceptive interference manipulations (a novel context, and a novel context plus a novel odor) can also be found in the five basic groups, as well as one compound interoceptive interference manipulation (a second tasteodor compound). Finally, one "unbiased" manipulation (a novel odor presented in a familiar chamber) was included. These five groups were found to provide the clearest picture of the results and were selected from the variety of groups as the experiments proceeded. They will be presented separately after the analysis and presentation of the data from all the groups in Experiments 1 8r 2. This is done to simplify the results. It is hoped that this will promote a general picture or feeling for the consistencies in the data as they are presented. These results can be considered as they are encountered, or referred to later for comparison purposes. M' ent One Experiment 1 was designed to establish the effects of different classes of 18 TABLE 1 INTE N R NETHRO H mess COMPOUND TASTE-ODOR INTERFERENCE ("INTEROCEPTIVE") 1.) A novel taste-odor compound (banana plus quinine) 02T2 group PLACE NOVELTY INTERFERENCE ("EXTEROCEPTIVE") 2.) A novel odor in a novel chamber CONTEXT-02 group some 3.) PLAIN WATER H20 group This is the comparison goup for interference effects 4.) UNBIASED SINGLE ODOR (banana) 02 group This is the comparison group for the compound interference manipulations with an odor component. 5.) PLACE NOVELTY WITHOUT ODOR CONTEXT group This is the comparison group for the CONTEXT-02 goup roman IOMNN 5 MIN DELAY noon 5 MIN DELAY ROOM RETURN HOME ozrz ODOR ’1 CONTEXT-02 PLUS 10 "IN 10 MIN H20 LICL TASTE #1 DELAY oz DELAY CONTEXT 19 interference manipulations interpolated between a target and delayed illness. In the absence of any sigtificant literature to guide selection of stimuli and conditions, we chose a wide variety of interference manipulations. Experiment 1 contained nine groups (see Table 2). Note that the labels used for the groups on all tables and graphs represent the interference manipulation presented to each goup. Two interoceptive manipulations were selected. One was a taste-odor compound, the 0212 g'oup, in which a .0004 M quinine solution was accompanied by banana extract applied to a disk which surrounded the water spout. This g'oup was run to determine the effect of a second taste-odor compound on the conditioning of the target taste-odor compound. The second interoceptive manipulation was a single stimulus consisting of just the taste component of the compound, the T2 group (the same .0004 M quinine solution), which was run as a comparison g'oup to determine the contribution of the taste component of the taste-odor compound to any interfering effects that compound might have. The indexing hypothesis would predict that these conditions would interfere with the conditioning of the target compound. It is possible that the 02T2 group, since it is a compound complex flavor that contains a second odor would show more interference than the T2 group. Four exteroceptive manipulations were included. Two were compound stimulus presentations, the SHOCK-02 g'oup (two 1 mA shocks were presented for one second in a familiar chamber 2.5 and 4.5 min into the 5 min session in Box 2, and were accompanied by banana odor applied to a disk around the water spout) and the CONTEXT-02 goup (placement in a novel context, a five gallon aquarium, that had banana odor applied to the disk around the drinking spout). These groups were run to determine the effect of placing an exteroceptive compound between the target taste-odor compound and illness. Shock was selected because its status as an external event and its COMPOUND TASTE-ODOR INTERFERENCE ("IN'I'EROCEP’ITVE") 1.) A novel tasteodor compound (banana plus quinine) om group PLACE NOVELTY INTERFERENCE ("EXTEROCEPTIVE") 2.) A novel odor in a novel chamber CONTEXT-02 group EXTERNAL CONDITIONING INTERFERENCE ("EXTEROCEPTIVE") 3.) Two novel odor and shock pairings SHOCK-02 group W 4.) PLAIN WATER 1120 group This is the comparison group for interference effects 5.) UNBIASED SINGLE ODOR (banana) 02 group This is the comparison goup for the compound interference manipulations with an odor component. 6.) BIASED SINGLE TASTE (quinine) T2 group This is the comparison group for the compound flavor interference manipulation. 7.) PLACE NOVELTY WITHOUT ODOR CONTEXT group This is the comparison g'oup for the CONTEXT-02 goup. 8.) SHOCK WITHOUT ODOR SHOCK group This is the comparison group for the SHOCK-02 group. 9.) NO TARGET FLAVOR COMPOUND W-OZT‘Z group This is the control group to determine the associability of the interfering flavor compound IOHI I 10H! ll 5 MIN MAY Rim 5 MIN DELAY ROOH RETURN HOME 02T2 CONTEXT-02 odor; r1 3’2: °2 10 MIN taste #1 10 MIN 02 LICL 12 DELAY DELAY CONTEXT SHOCK H20 v-02_T_2 21 association with exteroceptive stimuli (tones, odors, visual cues) is well established. Thus, an exteroceptive compound stimulus with known characteristics could be used as an exteroceptive interfering event. However, shock has reinforcing qualities as well (that is, shock is a salient US) so placement in an odorous novel context was added as a compound exteroceptive stimulus for comparison. If the indexing hypothesis is correct these groups would be expected to show little or no interference on the avoidance of the targets. The other two exteroceptive manipulations were single stimulus presentations, the SHOCK group (the same shock, with no accompanying odor) and the CONTEXT goup (placement in the novel compartment with no accompanying odor). These single stimulus groups were run to determine the contribution the odor component of the exteroceptive compound made to any interference effects that might be seen. An "unbiased" interference manipulation was included which consisted ofjustthebananaodoraroundthespoutinafamiliarchamber(the02 goup). We know that odor alone is not easily conditioned by illness (Rescorla 8r Durlach, 1981; Rusiniak et al., 1979), and that temporal contiguity is required for potentiation of odor by taste (Coburn et al., 1984; Holder 8r Garcia, 1987). Therefore, the prior presentation of the odor taste compound should not potentiate this odor, causing it to be configured into the target stimulus compound. Further, this condition should allow for an evaluation of the contribution of the odor component of the conditions in which an interfering stimulus compound (interoceptive or exteroceptive) was created by the addition of an odor. Another control condition provides the final group for this experiment, the W-02T2 group. This group was given only water in the first chamber and the "interference" taste-odor compound (.0004 M quinine solution accompanied by banana extract on a disk surrounding the drinking spout) in the second chamber. This group was run to assure the associability of the interfering interoceptive stimuli, and to test for the 22 strength of the avesion to the quinine-banana compound when followed 15 min later by illness. Kresge: Ninety six male Sprague Dawley rats (Holtzman Labs , Madison WI) wee assigred to nine groups and weighed and handled daily for 10 days before the experiment began. They wee adapted to shorter and shorter drinking periods until they were drinking only during one daily 10 min session. Water consumption was similar among animals and they were randomly assigied to experimental groups. For the next ten days the animals were run through the procedure they would be exposed to on conditioning days, and allowed to consume water, rather than test solutions, in the experimental rooms. This procedure consisted of first being placed in chambers withaccesstowaterforfivemin. ThechambersinRoom 1 were 21.5 cm X 23.5 cm X 21 cm high plexiglas boxes and were enclosed in sound-insulated boxes 73 cm X 38 cm X 47 cm high, two chambers to a box; the box was illuminated by a 15 W light with a smooth, quiet exhaust fan constantly running in each. They were then placed in individual stainless steel cages on a holding rack in a separate "delay" room where they spent 10 min The holding cages wee 17 cm X 24 cm X 18 cm high stainless steel rodent cages (Wahmon) with wire mesh floors. The rats were next put into chambersinRoomehichwere 19cm X24cmX21.5crnhighmodified Skinnerboxes; thesewereenclosedinsound insulatedboxesSl cmX94 cmX 47 cm high two chambers to an insulated box, in a third room where they wee again given access to water for 5 min. These boxes had a louder vibrating fan constantly running and were dim, being illuminated with only room light through a small 10 cm diameter hole in the ceiling. This second drinking episode was followed by another 10 min waiting period in individual cages in the "dela " room. The rats were then returned to their 23 home cages. Shock was delivered through the steel bar flooring in the conditioning chambers in Box 2 in the third room in the sequence. On conditioning days all rats, except for the W-02T2 group, were given 5 min access to the novel target taste-odor compound (0.1% saccharin in tap water and almond extract on a disk surrounding the water spout) in the first room and then, afte the 10 min delay spent in the holding room, were given 5 min access in a different room to the intefering events. The W—OZT‘2 group received wate in the first room and the quinine-banana compound in the second room. This was followed by a 10 min wait in the "delay" room where the rats were given an intragastric (ig) infusion of 0.15 M LiCl (2% body weight, or 20 ml / kg) before being returned to their home cages in the animal colony. The conditioning day was followed by two recovery days and then repeated. On recovery days the animals were run through the procedure with plain water. Two additional recovey days followed the second conditioning day, after which the animals were tested for consumption of each component of the two compounds, one each day for four consecutive days. Test days followed the same procedure as all the others. The tests of interest were for the odor and taste components of the target odor (OI-almond) and taste (Tl-saccharin); we wee primarily interested in the data from the target odor and taste tests. Data from tests with the interference stimuli are difficult to interpret because the interference stimuli were in the second position in the sequence (in Box 2), the animals received 5 min access to water in Box 1 prior to the test. Consequently, all consumption was very low, due to the satiation effect. Further, the stimuli presented in the second position were novel for some groups and not for others, so thee was differential neophobia. For these reasons the first test day was a test of the consumption of 01, and Test 2 was for consumption of T1, the target stimuli; data were then collected for 02 and T2. The data of interest for these stimuli are those from the W—OZTZ 24 group, who were rim to determine whether the interference stimuli were salient in this situation. This experiment was run in two replications. The first replication included six of the final nine groups (n's = 8 each). These groups included the HZO, 0212, T2, SHOCK-02, CONTEXT-02, and the 02 groups. The second replication duplicated the original 6 groups (n's = 4 each) and was run along with the three remaining groups, the CONTEXT, SHOCK and the W-02T2 groups (n's = 8 each), in orde to verify the effects of Experiment 1 and to add relevant controls. Refe again to Table 2 for a summary of all the groups in Experiment 1. Consumption was measured during habituation, conditioning and testing by recording the ml of fluid each animal consumed from a 50 ml graduated cylinder in each chamber. Thus, mean ml consumed is the variable recorded on all graphs, and the greater the amount of fluid consumed in the presence of experimental stimuli the less the aversion the animals demonstrated. Beating Figure 1 presents the results of Experiment 1 for the target odor and taste stimuli. When examining Figure 1 note that the HZO group, at the far left of the graph, is a control for the interference manipulations; it is a measure of the strength of the taste and taste-potentiated odor aversions when no additional interoceptive or exteroceptive events occurred in the second chamber, and the comparison groups for interference effects in the almond (01) and saccharin (T1) tests. Comparing the height of the other groups with that of the H20 group gives a measure of the interference effects; the higher the bar in relation to the HZO group the more the intervening event attenuated the conditioned aversion. On the day prior to the first conditioning trial the baseline water consumption was computed. For Box I MEAN ML CONSUMED MEAN ML CONSUMED 25 Almond Test (target odor) Exp. 1 BASELINE WATER CONSUMPTION: 10.85 ML 02 shock 2 02 context H20 shock context / no "Intere- "9 9'“ "exteroceptlve" Ini.”.f.nc. c.p“v." Interference Condition Saccharin test (target taste) Exp. 1 BASELINE WATER CONSUMPTION: 10.35 ML 3. § 3 c 8 N 0 00018 XI shock \\\\ III '.M x no ”Intero- "0 bl“ "sxterocsptlve" Interference “pun" Interference Condition 26 it was 10.85 ml, and for Box two was 6.0 ml. Water baselines were not different among treatment groups. Neophobia measures - consumption of test fluids by animals not presented with these stimuli during conditioning - are not presented in ”I68 figures. Neophobia can show complicated non-associative effects, especially after toxiphobia conditioning. Note will be made in cases where differential neophobia effects occur. A 9 X 2 mixed ANOVA was run on these data, the between subjects factor was interference manipulations (the nine groups) and the within factor was target (01 or T1). This analysis revealed significant main effects for interference conditions F(8,87) = 4.79, p < .01 (MSE = 9.46), type of target (the two components of the target compound), F(1,87) = 16.60, p < .01 (MSE = 4.72), and a marginal interaction F(8,87) = 1.99, p = .059 (MSE = 9.39). This analysis demonstrates that the overall consumption of the target odor and taste were significantly different. Mean consumption for the target taste (saccharin, T1) was 3.25 ml, while the mean consumption for the odor component (almond, 01) was 1.9 ml. It is often the case that the potentiated odor aversion is stronger than the aversion for the taste. It is possible that the status of odor as a distal cue is responsible, since the animal does not need to touch the spout to identify its contents, but the circumstances that promote this differential aversion in some cases but not in others are currently unclear. The results for the interaction (32 = .059) suggested a reliable difference between interference manipulations depending on the target. Given this marginal interaction, simple effects tests were run for the two components of the target and they revealed the following. There was a significant difference among groups on the almond odor tests, F(8,87) = 5.92, p < .01 (MSE = 6.87); Duncan's Post hoc comparisons showed that the consumption measure for the OZ-SHOCK group was reliably different from the water control (HZO), indicating interference on the target odor. All the groups were reliably different from the W-O2T2 group (which was exhibiting 27 neophobia, since these subjects had never encountered the almond odor before), indicating that an odor aversion was shown by all the experimental groups. There was not a significant interference effect for saccharin (the target taste) F(8,87) = 1.93, 2 >05 (MSE = 7.31) showing that there were no reliable differences in consumption between interference conditions in the taste condition. Simple effect tests also revealed that the H20 and 0212 groups differed across targets, F(1,87) = 11.76 p < .01 (MSE = 4.72), and F(1,87) = 7.68 p < .01 (MSE = 4.72), respectively. In both cases this reflects lower consumption on the odor test than on the taste test (mean consumption = 0.04 ml for almond, 3.1 ml for saccharin), and in the case of the 0212 group it reflects the increased interference by 0212 on the taste rather than the odor component (mean consumption = 0.96 ml for almond, 3.42 ml for saccharin). Figure 2 shows the data for the interference stimuli, 02 and 12. Only the groups which had either banana or quinine followed by illness are shown since these are the the aversion conditions. Also recall that these subjects had received 5 min exposure to water in Box 1 just before the test so consumption is generally depressed due to satiation. Again, neophobia measures are not represented in this figure. The W-021‘2 group gives a measure of the strength of the aversions to the second taste and its potentiated odor when no additional interoceptive event occurred in the first chamber. A one way ANOVA was run on the consumption data for banana (the interfering odor): F(8,87) = 3.68,}; < .05. The data from the neophobia groups (those groups for whom banana was not a component of the experimental situation) were pooled and comparisons among means were computed. These analyses showed that all the groups that had experienced the interfering odor except the CONTEXT-02 group showed a significant difference in consumption from those groups which were not presented with this stimulus during conditioning. The results are as follows: 02, F(1,87) = 6.9, ; 0212, F(1,87) = 8.2 , MEAN ML CONSUMED (PLUS 0.1) MEAN ML CONSUMED 28 Banana test (interference odor) Exp. 1 BASELINE WATER CONSUMPTION: 6.0 M. (2T2 W42 SHOCKOZ W 02T2 771'.’/’/" xteroceptlve" control "Interoceptlve" Interference Condition QUlNlNE TEST (INTERFERING TASTE) Exp. 1 BASELINE WATER CONSUMPTION: 6.0 ML ”Interoceptlve" "exteroceptive" control Interference Condition 29 W-02'1‘2; F(1,87) = 5.49; SHOCK-02, F (1,87) = 8.9; MSE = 2.24, p < .05 in all cases. The elevated consumption by the CONTEXT-02 occurred because the CONTEXT-02 subjects were being tested for an aversion to this odor in a chamber different from the one in which they had experienced it during conditioning. After the testing was completed these subjects were tested in the aquarium in which they had experienced the odor during conditioning and were found to avoid the odor completely (mean consumption = 0.0 ml). The previous analyses demonstrate that all subjects that experienced the odor in the second chamber acquired an aversion to it, regardless of how many other internal or external events they learned, and that the odor component of the interfering compound supported an odor aversion (mean consumption for the W-021'2 group a 0.00) within the present experimental parameters. A one way ANOVA was run on the data for the quinine test (interfering taste): F(8,87) = 4.6 p < .01 (MSE = 207). All the groups which had not been exposed to quinine (neophobia measures) were pooled and comparisons among means were computed for: 1.) the W-021'2 group, the control for an aversion for quinine F(1,87) = 4.27, p < .05 (MSE = 207), which demonstrated that the conditioned aversion for quinine was significantly stronger than the neophobia measures 2.) for the 0212 group, F(1,87) = 7.74, p < .01 (MSE = 2.07), which showed that the animals could learn about all the stimuli and support two successive aversions, and 3.) the 12 group, F(1,87) = 11.49, p < .001 (MSE = 2.07), demonstrating that the odor component of the compound was not required for the ability of the animals to learn aversions to two flavors (actually, two flavors and one odor) presented in sequence. 12 . . n The major conclusion that can be reached from the results of Experiment 1 is that the association between a target taste and its potentiated odor is not 3O easily disrupted by either the exteroceptive or interoceptive events that we presented during the CS-US interval. In the present situation only the compound exteroceptive manipulation, the SHOCK-02 condition produced statistically significant interference. These results are directly opposite to the predictions of the indexing hypothesis, which predicts interference effects only by interoceptive events. It is interesting that neither of the single stimulus exteroceptive manipulations (SHOCK and CONTEXT) produced interference. Apparently, in this situation either the addition of an odor cue to these conditions, or the fact that they were compound cues, was necessary to produce the interference effects. Since the control group (group P120) demonstrated the expected suppression of consumption of the taste and odor components of the target we have a basis for confidence in these results. However, overall the aversions were quite strong and it is possible that this floor effect obscured the effects of the various interference manipulations. Figure 3 shows a subset from the same data, the basic five groups. A 2 X 5 mixed ANOVA was run on these data, with the between subjects factor being interference manipulation (the five basic groups) and the within subjects factor being target (01 or 11). This analysis revealed no significant main effect for interference condition, F(4,51) = 1.59, p_ > .05 (MSE = 8.31 ), a significant effect for type of target (the two components of the compound), F(1,51) = 33.45, p < .01 (MSE = 3.4), and a nonsignificant interaction 2 > .05. The removal of the high consumption of the W-021‘2 group, which had not been conditioned to avoid the target compound, is largely responsible for the lack of a main effect for interference condition. This attests to the presence of a floor effect, particularly on the odor component. This analysis shows that overall consumption of the target odor and taste were significantly different. Mean consumption for the target saccharin taste (11) was 3.20 ml, while the mean consumption for the almond odor (01) MEAN ML CONSUMED MEAN ML CONSUMED 31 a- Almond test (target odor) Exp. 1 BASELINE WATER CONSUMPTION: 10.85 ML 6-4 1 a: 2 4< g N O 2- 8 O S O U N I o n "IMII'O- "° bl“ "exteroceptlve" Interference ceptIve" Interference CondItIon 31 Saccharin test (target taste) Exp 1 BASELINE WATER CONSUMPTION: 10.85 ML 3: 6- a . a 8 o 8 4.. o E 8 g E O 0 2-4 0 , 0° "Intero- "0 bl“ "exteroceptlve" Interference ceptlve" Interference Condition 32 was 1.15 ml. The analysis also suggests that there were no significant interference effects because of the overall strength of the aversions. Again, the control group (H20) demonstrated typical results, with consumption in the presence of the odor lower than in the presence of taste. The context plus odor manipulation did not show a significant difference from the control group here when only these 5 groups are analyzed, even though it was not significantly different from the SHOCK-02 group in the previous analysis, in which the SHOCK-02 group showed a significant effect. t Tw A small but reliable difference among interference manipulations on a potentiated odor aversion in Experiment 1 was demonstrated. However the interference effect was produced by an exteroceptive compound manipulation (SHOCK-02), and not by an interoceptive compound (0212). The strength of the taste—potentiated aversions across all the manipulations may have obscured interference effects. Since floor effects may have been present, it seemed appropriate to use a weaker conditioning arrangement to unmask interference effects. There are a number of ways to weaken a conditioned aversion, including preexposure to the elements of the compounds, lower doses of LiCl, less intense flavor stimuli, and fewer conditioning trials. Each of these methods comes with its unique difficulties. Reducing the number of conditioning days was chosen since a great deal of the research using taste aversion methodology commonly uses either one or two conditioning days. Reducing the number of conditioning days should weaken conditioning without any differential effects as can result with preexposure, or by altering the salience of the stimuli by changing the taste concentration. Furthermore, Sjoden and Archer (1989) have recently suggested that number of conditioning trials may be an important variable in its own right. They argue that all that is to be 33 learned about the flavor stimuli occurs on the first trial, and thereafter rats learn about the environmental stimuli in the experimental situation. Their analysis was not extended to the components of the flavor stimuli, but it is conceivable that in complex situations such as the present one, learning about primary gustatory stimuli and learning about other aspects of the conditioning environment occur along different time frames. If this is the case it is possible that learning about the exteroceptive cues interfered with learning about the target stimuli on the second trial, since contextual learning isthoughttobeincremental. Notethatthisanalysisisnotinlinewith the implicit assumption of the indexing hypothesis, which suggests that taste ought to protect odor from interference during each exposure and that such learning should be independent of exteroceptive learning. However, the . present experiment should provide an evaluation of this suggestion. If Sjoden and Archer are correct, then interference in the present experiment should be stronger by interoceptive than by exteroceptive manipulations since the subjects will not have had the second conditioning session. Alternatively, if interoceptive stimuli are incapable of interfering with each other and are simply also associated with illness, there should be no interference effects at all, since their suggestion implies that attention is directed to consumatory stimuli on the first trial and only spreads to exteroceptive stimuli on the second trial. Another issue addressed in the present experiment was that of the contribution of the odor to the exteroceptive compound interference manipulation. Although the SHOCK-02 was the only group that showed significant attenuation of the avoidance to the target odor, the CONTEXT-02 was within 0.5 ml of it. The context and shock alone groups showed consumption equivalent to the interoceptive manipulations, that is, five groups (three interoceptive and two exteroceptive) showed consumption measures that were halfway between the control and the interfering stimuli, 34 and none of these was an exteroceptive cue plus odor condition. Therefore a set of conditions which produced exteroceptive associations not involving odor were included. White noise and white noise paired with shock were added to this experiment as single and compound exteroceptive conditions that did not contain an odor in order to determine whether it was specifically the addition of an odor to the exteroceptive manipulation or the fact that it was an exteroceptive compound that caused the interference found in Experiment 1. It seemed unlikely that the latter of these two possibilities was the case since it made no difference whether the cue was single or in compound in the case of the interoceptive manipulations (the O2, 12 and 0112 groups showed little variation). However, given that the exteroceptive and interoceptive systems may function under different rules, it seemed advisable to include such a manipulation. The 12 group, which did not differ from either the 02 or 0212 groups was dropped from the experiment since it did not contribute additional information. Afinalissueofimportanceinthenextofthisseriesofstudies was based on observation of behavior in the delay room during the running of Experiment 1. Beginning on about day 14 of the 21 day schedule of Experiment 1,all of the rats began to behave in an agitated, active and aggressive way during the two 10 min waiting periods in the delay room. They flung themselves against the walls, chewed the wire cages, leaped up and about, and a few of them began to nip the experimenter regularly (mean bites = 4.5 per day). Since this behavior began at the same time as the first conditioning day, it was unclear if this was a result of the long habituation to the predictable schedule of imposed delays between drinking bouts, or whether it was the conditioning episodes that set it off. Therefore in the present experiment the habituation period was shortened to six rather than 10 days. A difference of four habituation days is unlikely to make a difference in the results since stable water drinking has been achieved by this time, but it 35 would be expected to affect learned tinting effects and adjunctive behavior (Staddon Gt Simmelhag, 1971). If the behavior is seen in this experiment as well it can be concluded that it was not the long habituation period that caused it but the onset of conditioning. The emergence of adjunctive behavior indicates that the delay room was not a place for the animals to simply wait as we had envisioned it, but a place of some importance that could contribute to the interference effects in some groups and not in others. If this is the case these additional important events (two placements in the delay room), and the nonspecific stress they produced, could interact with the experimental manipulations. Refer to Table 3 for a summary of the experimental groups in this experiment. mgr: Ten groups of animals were run identically to those in Experiment 1 except for the following changes: 1.) The 10 days of habituation to the conditioning chambers was shortened to six days. This was the approximate number of days it took to reach steady water intake and allowed the conditioning trial to occur on Day 7, several days before the adjunctive behavior was observed in the delay room in Experiment 1. 2.) One conditioning trial was employed rather than two. It was expected that this would weaken conditioning and thereby allow a finer analysis of the effects of the interfering manipulations on the conditioning of the components of the target 0111 compound. 3.) Two exteroceptive manipulations were added. Since 02 was a component of each of the compound exteroceptive stimuli, it was unknown whether the interference produced in these conditions (CONTEXT-02 & SHOCK-02) was a result of some special property the odor brought to the exteroceptive compounds or was simply the result of the fact that it was a COMPOUND TASTE-ODOR INTERFERENCE ("INTEROCEP'I'IVE") 1.) A novel tasteodor compound (banana plus quinine) 0212 group PLACE NOVELTY INTERFERENCE ("EXTEROCEPTIVE") 2.) A novel odor in a novel chamber CONTEXT-02 group EXTERNAL CONDITIONING INTERFERENCE ("EXTEROCEPTIVE") 3.) Two novel odor and shock pairings SHOCK-02 group EXTERNAL COMPOUND INTERFERENCE ("EXTEROCEP'I'IVE") 4.) Two pairing of white noise and shock NOISE-SHOCK group W 5.) PLAIN WATER H20 group This is the comparison group for interference effects 6.) UNBIASED SINGLE ODOR (banana) 02 group This is the comparison group for the compound interference manipulations with an odor component. 7.) PLACE NOVELTY WITHOUT ODOR CONTEXT group This is the comparison group for the CONTEXT-O2 group. 8.) SHOCK WITHOUT ODOR SHOCK group This is the comparison group for the SHOCK-02 group. 9.) NOISE WITHOUT SHOCK NOISE group This is the comparison group for the NOISE-SHOCK group 10.) NO TARGET FLAVOR COMPOUND W-OZTZ group This is the control group to determine the associability of the interfering flavor compound IOHII Ioulu 5mm owner SHIN cannon ammm 02T2 comm-02 SHOCK-02 NOISE-SHOCK “”1" IO MIN H20 IO MIN taste #1 02 “CL DELAY cou'rgx'r DELAY SHOCK NOISE T 37 stimulus compound. Therefore a noise-alone group and a noise-shock group were added to this experiment. These stimuli were presented in the following way: For the noise-alone group a 85 dB white noise was presented for 30 seconds at 2.5 and 4.5 seconds into the second water exposure in Box 2. The noise-shock group was heated the same except that each presentation of the white noise was followed by a 1 mA shock of 1 second duration. If the exteroceptive compound manipulations interfered with the target aversions because of the odor component of the compound, then the noise-shock group should not show comparable interference; however, if the interference effects were the result of the animals having learned an additional association in a different system between the presentation of the target and illness then this group should show similar interference of the target associations. The noise alone group is a comparison for the context alone, odor alone and noise-shock groups; evaluation of the relative interference effects between these stimuli is a contribution towards determining where odor lies in the external-internal distinction, at least in this paradigm, and evaluating any contribution the noise might make to the noise-shock group. If the only exteroceptive manipulations that interfere with the target CS are those accompanied by an odor, odor may have special properties not accounted for by the indexing hypothesis. This experiment was run in two replications. Since there were now 10 groups of eight animals each and the experimental procedure was lengthy, it was not desirable to run them all at once. By the time the first half of the animals had been run the second half would be nearing quite a different time of day and state of deprivation. baseline water consumption for Box 1 was 10.2 ml, and for Box 2 was 5.54 ml. 38 Beside . TheresultsofthetestswiththetargetsOl andT1inExperiment2are shown in Figure 4. A 2 X 10 mixed ANOVA was run on the ml consumed in which the between groups factor was interference condition and the within subjects factor was type of target (odor or taste). This analyses yielded a significant main effect for type of target, F(1,70) = 6.0, p < .01 (MSE = 8.35), interference condition F(9,70) = 257, p < .01 (MSE = 4.59) and a significant interaction, F (9,70) = 22 (MSE = 4.59). Simple effects tests revealed significant differences among interference conditions on the target odor, F(9,129) = 3.56, p < .01 (MSE = 6.46), but not on the target taste, F(9,129) = 1.31, p > .05 (MSE = 6.46). Unlike Experiment 1 the size of the interference effect by interoceptive and exteroceptive stimuli on the odor and taste components was similar, but there were several groups that interfered more with the odor component than the taste component. Furthermore, the strength of the odor and taste aversions in the control group (HZO) were of the same magnitude, and there was also not much difference between the total consumption in the odor and taste tests. Unlike Experiment 1, in which the mean overall saccharin consumption (3.25 ml) was almost double that of almond (1.95 ml), the mean consumption in the present experiment for the almond test was 2.9 ml while the mean ml consumption for the saccharin test was 2.0 ml. It is likely that the significant main effect for target and the significant simple effect for groups in the almond test was carried largely by the low 01 consumption by the CONTEXT group; there was no such facilitating effect for the CONTEXT group on the taste component of the target compound. Simple effects tests for the two target stimuli revealed a significant effect for the odor component, F (9,129) = 3.56, p < .01 (MSE = 6.46), but not for the taste component. Duncan‘s Post hoc analyses for the odor test MEAN ML CONSUMED MEAN ML CONSUMED 39 Almond test (target odor) Exp. BASELINE WATER CONSUMPnON: 10.2 ML :2 2 C 8 ,N 0 no 'i"t°f°‘ "° 'extrcceptive' interference ceptIve' bias Interference Condition 02 shock noise-shock noise x O O .C W - - - — _ — — — — — — — — _ - _ — _ _ _ — — — — II|||||lllllIlllllllllllIlllIlIllIlIHlllIIIIIIIIIII 02 Saccharine Test (target taste) Exp. 2 BASELINE WATER CONSUMPTION: 10.2 ML 3 a N O noise-shock context IIIIIIIIIIIIIIIII “0°" noise 02 context \ . "0 '"Intero- 0° 'extroceptive' Interference ceptIve' bias Interference Condition 40 (p < .05 in all cases) showed that the control group (HZO) differed from the 02, 0212, CONTEXT-02, SHOCK and the W-0212 (a neophobia measure) groups, which indicates attenuation of the target potentiated odor aversion in these groups. That is, these four groups showed interference effects on the odor aversion. The CONTEXT group, which showed lower consumption than the control group (HZO), also differed statistically from these groups. N 0 other significant differences were found. The NOISE-SHOCK group did not differ from the SHOCK-02 group or the NOISE or SHOCK groups suggesting that it was not simply the addition of an odor to the exteroceptive cue that caused interference. Thus, Experiment 2 replicates the interference effects shown by exteroceptive cues in Experiment 1; and in addition there was interference by both the compound flavor group (0212) and the odor group (02) as well. In summary, two of the exteroceptive conditions (SHOCK & . CONTEXT-02), the unbiased group (02), and the compound interoceptive manipulation (0212) were found to provide interference on the odor component of the compound in this experiment. Because there was a significant main effect for target but the simple effects test for T1 was not significant a Duncans Post hoc analysis was run for the data collapsed across targets; these same four conditions were found to differ significantly (p < .05). An interesting result of this experiment is found in the simple effect test for the control group (HZO) across targets F(1,70) = 0.95, p > .05 (MSE = 4.57). A glance at the figures will confirm that there is no difference between the consumption of the test stimuli for the control group (mean ml consumed for 01 = 0.65, and for T1 = 0.72 ml). It is often found that the potentiated odor aversion is stronger than the aversion to the potentiating taste (Rusiniak et al., 1979). It is possible that the increased avoidance of odor stimuli in the potentiation preparation is a result of the dual exposure of the subject to odor but not to taste. That is, an animal can avoid tasting the aversive stimulus on the second conditioning day but it cannot avoid the odor, and this results in 41 differential conditioning. Also note that the groups that showed significant differences in consumptionbetween targets allconsumed moreinthepresenceofOl than T1. So despite the equal aversions for the targets by the control group, there were bigger interference effects on odor than on taste when an interference treatment was given. Simple effect tests for target at the following groups were found to be significant: CONTEXT-02, [F(1,70) = 6.12, p < .05 (MSE = 4.57) mean consumption of 01 = 3.75 ml & mean consumption of T1= 1.09 ml], 0212, [F(1,70) = 5.75, p < .05 (MSE = 4.57) mean consumption of 01 = 4.43 ml at mean consumption of T1= 1.87 ml], & W-0212 (showing neophobia), [F(1,70) = 9.42, p < .05 (MSE = 4.57) mean consumption of 01 = 5.31 ml & mean consumption of T12 2.03 ml]. In summary, there were larger interference effects on odor for some groups, and these effects were in the same direction as those from Experiment 1. The overall consumption measures went in the opposite direction, with consumption during the odor test slightly highg than during the taste test. In Experiment 1, overall consumption of the taste target was higher. However, it is important to keep in mind that while overall consumption of the taste component was significantly higher in Experiment 1, there were no interference effects on the taste component relative to the control group (HZO) in that experiment. In the present experiment the overall consumption measures were more similar between targets but consumption of these targets by the control groups was comparable, as were the interference effects. This suggests that the reduction of conditioning days had the effect of reducing the strength of the odor aversion and increasing the strength of the taste aversion, as measured by the control group (H20). The interference effects on the taste target were larger but no more specific than in Experiment 1, however this increase in interference was a result of the decrease in consumption by the control group, not an increase in magnitude of the 42 disruption shown by the interference conditions. Further, in this experiment the interference on the odor component increased to include interoceptive as well as exteroceptive cues, and resembled the interference effects on the taste component. Finally, there was no distinction between single and compound exteroceptive cues as there had been in Experiment 1. Figure 5 shows the data for the interference stimuli, 02 and 12. Again, neophobia measures are not shown for simplicity. The W-O212 group gives a measure of the strength of the aversions to the second taste and its potentiated odor when no additional interoceptive event occurred in the first chamber. No significant differences were found among groups or between targets for the interference stimuli, p > .05. That is, all the subjects that had been exposed to the second odor cue avoided it in testing and there were no differences among them. 2mm Reducing the number of conditioning trials from two to one increased the interference effects although it also resulted in both exteroceptive and interoceptive manipulations attenuating the avoidance response to the odor component of the target. Further, the single stimulus exteroceptive cues did not show a marked difference from the compound exteroceptive cues. These differences cannot be attributed to the removal of whatever stressor caused the adjunctive behavior in the delay room in Experiment 1 since the same behavior was observed in the present experiment. Clearly, a change in the experimental procedure will be required to banish whatever contribution this stress makes to the interference effects observed so far. Finally, the CONTEXT group showed an increased avoidance of the odor cue relative to the attenuated avoidance shown by all the other interference conditions. Noting the possibility of spurious results when so many conditions are examined, this "potentiation" of the already potentiated MEAN ML CONSUMED (PLUS 0.1) MEAN ML CONSUMED 43 BANANA (INTERFERINC coon) Exp. 2 BASELINE WATER CONSUMPTION: 5.54 ML E W-02T2 m ”Interoceptlve Interference Condition QUININE TEST (INTERFERING TASTE) Exp. BASELINE WATER CONSUMPTION: 5.54 ML eroceptlve" control 2 5 u 4 .. 3 - 2 n N C 1 - '8 8 Z 0 ”Interoceptlve" "exteroceptive" Interference Condition 44 odor association is interesting and will require further investigation. It is possible that the contextual change, since it had no novel odor to make it a salient new location, served to "mark” (Lett, 1975) the target stimulus event or "bridge" (Kaplan & Hearst, 1982) the delay between that compound taste-odor presentation and illness. Figure 6 shows the subset of data for the target taste and odor tests. The mixed 2 X 5 AN OVA (target by groups) revealed a significant result for groups, F(4,35) = 4.97, p < .01, a nonsignificant effect for target F(1,35) = 3.4, p > .05, and a significant interaction, F(4,35) = 299, p_ < .05. Simple effects tests showed that there was a significant effect for groups on the target odor, (01) F(4,66) = 6.39, p < .01, but not for the taste component ('11), F(4,66) = 2.08, p > .05. Visual inspection of the data for the taste stimuli suggest that the interference on this stimulus was significant. If the simple effects tests are calculated using the error term for one cell (Kepple, 1982, pp. 442-446) rather than the within groups error term (Weiner,1962, pp. 529-532) the tests show a significant effect for the taste stimuli F(4,35) = 3.35, p < .05. Given the tremendous variance in the 01 data relative to the small variance in the taste test, it seemed inappropriate to include this variance in the analysis for the taste component. Duncan's Post hoc analysis revealed that the 02 group differed from all the other groups and that there were no other differences. Duncan's Post hoc analyses on the 01 test revealed that the CONTEXT and P120 groups differed from all the others (02, 0212, and CONTEXT-02), which did not differ from each other. This indicates that all the interference manipulations interfered with the avoidance of the odor component of the target, relative to the control (1-120) and CONTEXT groups. These analyses again show that both interoceptive and exteroceptive manipulations interfered with target odor conditioning in this experiment. Simple effects tests for targets at groups showed that the CONTEXT-02 MEAN ML CONSUMED MEAN ML CONSUMED 45 Almond test (target odor) Exp. 2 BASELINE WATER CONSUMPTION: 10.2 ML 0212 I 02 context lIlIIIIlIIIIIIIIIlIIIIlIlIIlIIIIIIIIIIIIIIIIIIII 02 no "lntero- no 5'" "exteroceptive" Interference ceptIve" ‘Interference Condition Saccharin test (target taste) Exp 2 BASELINE WATER CONSUMPTION: 10.2 ML 02 context 8 no "lntero- “0 5‘“ "exteroceptive" Interference ceptIve” Interference Condition 4e and 0212 groups differed between targets [F(1,35) = 6.98, p < .05, and F(1,35) = 6.5, p < .05 respectively, (MSE = 4.04)]. Both these groups showed higher consumption (more interference) of the odor component of the compound than the taste component. Interim summary Experiment 1 was designed to assess the interference effects of interoceptive and exteroceptive events interpolated between the presentation of a target taste-odor compound and illness on the conditioning of the target. The two days of conditioning took place in two separate familiar environments, with placement in familiar cages during the delay. Contrary to predictions based on the indexing hypothesis, it was found that an exteroceptive compound stimulus consisting of a second odor paired with shock interfered with avoidance of the odor component of the target in testing, and minor (nonsignificant) interference occurred as a result of interoceptive manipulations. While interference by the compound exteroceptive manipulation was statistically significant, it appeared as a minor modification of the overall strong aversion. It was therefore desirable to attempt to weaken this aversion and examine the relative effects of the interference manipulation against a less suppressed overall intake. Experiment 2 assessed the effect of reducing the conditioning days from two to one on eight of the original groups from Experiment 1 and added two groups (NOISE, and SHOCK-NOISE) . These groups were designed to evaluate whether the interfering effect of the compound exteroceptive cue manipulation in Experiment 1, the only interference condition that showed a disruptive effect, was a result of some property of the odor component of the compound or simply the learning of a second, exteroceptive association. The results of this experiment revealed that both interoceptive and exteroceptive interference manipulations were effective in producing disruption. 47 Further, the clear diffeerces in interference effects between single and compound exteroceptive manipulations, so striking in Experiment 1, was not evidert. Rather, there was very gereral interfererce on both classes of inte'ference conditions. While some groups showed larger interfeence effects on odor, the size of interfererce effects and the total overall disruption of the taste and odor componetts of the targets were similar. In addition, unlike Expeiment 1, the taste and odor aversions evidetced by the control group wee equal for the two componerts of the target. W Among the several diffeences between Experimert 1 6r 2 the number of trials seemed most likely to influerce aversion learning. Therefore this expeiment was designed to determine whether the differences in interfeenceeffectsinExperiments 1 &2wereduetothedifferetcesinthe numbe' of exposures to the stimuli, i.e. training trials, or to a difference in associative strength. The results from Expeimert 1 provided evidence that contradicted the indexing hypothesis' assumption of system indeperdence in that exteroceptive stimuli accompanied by an odor disrupted conditioning to both components of a target odor taste compound. The results of Expe'iment 2, in which one rather than two conditioning days was employed, showed inte'ference of the target by both interoceptive and exteroceptive events, and the pattern of interfeence on the taste and odor components was very differert than that for Experiment 1. However, the question remained as to whether these diffeential interference effects were a result of the different numberofexposurestothestimuli,orthedifferercesinthesizeofthe conditioned aversion. In addition to providing different levels of conditioning (magnitude), two trials affords two separate opportunities for taste and odor to interact (frequency), and in the case of the present series of expeiments, for the consequences and inte'fering stimuli to interact with the target odor and taste. 48 The analysis of one versus multiple trials has been virtually ignored in the literature (Sjoden Gr Archer, 1989), although it has long been suspected by researchers in the field that there are qualitative differences in the mechanisms responsible for one and multiple trial learning (Bouton, M., personal communication, Boston MA, April 1, 1989; Westbrook, Homewood, Horn, 8: Clarke, 1983). However, if Sjoden and Archer (1989) are correct in their suggestion that in taste aversion paradigms learning about interoceptive stimuli occurs in one trial and that thereafter learning concens exteroceptive events, we should have seen little interference by exteroceptive events in Experimert 2. Recall that the results of Experiment 2 showed inte'ference by both classes of events rather than the elimination of exteroceptive interference. However, if the differential effects seen in the two previous experiments were the result of the attenuation of the strength of the association, we would have expected to see an increase in interference in the same groups that showed interference in Experiment 1 and perhaps the emergence of interference in groups in which we were unable to detect it in Expeiment 1. Instead, the interference, rather than coming from only exteroceptive manipulations, was equal between the two classes of stimuli. An issue that had not yet been successfully addressed in these experiments was that of the adjunctive behavior seen in the delay room. This behavior was still in eviderce in Experiment 2, suggesting that reducing the habituation period was not effective in eliminating it. Therefore a new procedure, developed during pilot work on a similar problem, was instituted for this experiment. This procedure eliminated the delay rooms altogether, and familiarized the subjects to the experimental rooms (chambers, boxes, etc.) on separate days so that the rooms and conditioning chambers were not discriminative stimuli for later everts. This procedure was found to eliminate the adjunctive behavior altogether. An additional advantage was that it allowed us to assess the strength of the avoidance to the interference 49 cues without confounding the measures with satiation effects, so we could assess aversions to the interfererce cues with more ce-tainty. Esteem Experimert 3 was subdivided into Expeimerts 3A a: B, and directly compared high and low associative strergth across two conditioning trials. This was achieved by manipulating the dose of LiCl. The intensity of the US remained at the level of the previous expeiments for half the subjects and was reduced for the other half of the subjects. The intensity of the LiCl was reduced while still preserting the subjects with two conditioning trials making it possible to contrast the results of the Low dose experiment (3B) with the results of Expeiment 2 (one trial) to determine whether the difference in the results betweer Experiments 1 8r 2 were due to differential exposure or weaker associative strength. For the High dose groups (Experiment 3A) the dose of LiCl was the same as in Experiments 1 & 2, (0.15 M) and for the Low dose groups (Expeiment 3B) the LiCl was half that dose (0.075 M). This equated the total amount of LiCl for the Low dose groups with the amount of LiCl that was administered on the one conditioning trial in Experiment 2. It should be noted that the high dose of LiCl (127 mg/ kg), referred to as strong conditioning, is the modal dose for taste aversion work. This dose of LiCl does not render the animals flaccid and observation suggests that they have completely recoveed from its effects within about two hours. A low dose of LiCl (63.5 mg/ kg) has been shown to produce attenuated taste avoidance (N achman 8r Ashe, 1973), and its effects are not easily detected by the animals' behavior. By refering to these conditions as strong and weak we do not mean to imply that the high dose condition is unusually severe. In fact, in a discrimination study, our low dose was found to be close to threshold for animals who were self administering lithium (Rusiniak, Garcia, .1: Hankins, 1976). 50 Five groups (n's = 8 each; total N = 40) were run for each expeiment. They consisted of the five "basic" groups that were preserted separately after each of the previous experimerts and are listed in Table 1. They included one basic control, the target taste-odor compound followed by no interference manipulation (1120), to which all the other manipulations can be compared to dete'mine the extent of the interfeence they produce. Two exteroceptive inteference manipulations (a novel context, CONTEXT, and a novel context plus a novel odor, CONTEXT-02) and one compound interoceptive interference manipulation, a second taste-odor compound (0212). Finally, one "unbiased" manipulation, the novel banana odor presented in a familiar chamber, (02) was included. The experiment was run in two replications. Half of the subjects for each group (n's = 4; total N a 40) of both Experiments 3A & B wee run for each of the replications. Habituation trials Animals were habituated to the chambers in a 3—day cycle: Box 1 (habituation day 1), Box 2 (habituation day 2), and then Box 1 - home cage - I” Box 2 - home cage (habituation day 3). All drinking sessions remained at 5 min. This was repeated three times, for a total of eight habituation days, with the third cycle culminating in the first Conditioning trial. This was followed by two recovery days, Box 1 for the first recovery day and Box 2 for the second, and followed by the second conditioning trial. Two recovery days followed in the same fashion as the previous recovery days. Consequently, rather than habituating animals to a daily three-room sequence of Box 1 - delay room - Box 2 — delay room, in the entire expeiment there were only four days when rats were placed in two rooms on the same day. Four test days were then conducted. Each test was concluded in the chamber in which that component had been conditioned, and each component of the compound was evaluated singly and on a separate day. Since subjects were only tested in a single 5 min test on each day, this procedure allowed direct comparison of the strength of 51 the aversions to all the components of both the target and interfeence stimulus with equal deprivation levels. All othe' expeimental parametes remainedthesameasinExpeimentsI 8:2. If the results of Expeiment 3A with the high dose of LiCl, on two training trials resemble those of Expeth 1, which used a high dose and two days of conditioning, and the results of Experiment 38 with the low dose of LiCl look like those of Expeiment 2, a high dose and one day of conditioning, then we have evidence that the difference in the two previous experiments was due to the strength of the association and not some differential effect of number of exposures. If the pattern of results for the two parts of Experiment 3 look similar to each other and to Experiment 1 then we can assume that the effects seen in Expeiment 2 were due to the single conditioning exposure. 11.6% The target sfimuli (91 g T1) The results for the target stimuli for Experiments 3A 8: B are shown in Figures 7 8r 8. I will describe each of the studies separately, and compare them with the results of Experiments 1 6t 2 before proceeding to the statistical analysis. The baseline water consumption for Experiment 3A, the high dose condition, was 9.1 ml in Box 1, and 8.9 ml in Box 2; for Experiment 3B ,the low dose condition, the baseline water consumption was 9.0 ml in Box 1, and 8.7 ml in Box 2. Figure 7 shows the results for the target stimuli for Experiment 3A, which used the high dose of LiCl. The control group results showed that there was a stronger odor than taste aversion. There were also interference effects on the odor aversion that were larger than those on taste. Some interference on the target odor was shown by groups from both the interoceptive and exteroceptive manipulations, although significant interference was shown only by the interoceptive compound interference MEAN ML CONSUMED MEAN ML CONSUMED (PLUS 0.1) 52 a] ' Almond test (target odor) Exp. 3A BASELINE WATER CONSUMPTION: 9.1 ML 6.: 4‘ E 3 ° 8 0 2« E C a 3 8 I 8 0 _ "0 "lntero- no blee “exteroceptlve” Interference ceptlven Interference Condition Saccharln test (target taste) Exp. 3A BASELINE WATER CONSUMPTION: 9.1 ML H20 0212 02 context context no "Intero- no bl“ "exteroceptlve" Interference “WIN." Interference Condition 53 manipulation (0212). The aversion to the taste component was less pronounced and there were no interference effects demonstrated. Figure 8 shows the results for the target stimuli for Experimert 3B, which used half the dose of LiCl. There was a pattern of interference on the odor aversion that was similar to that found in Experiment 3A, with significant interference shown by the 0212 group. On the taste component the aversion was almost as strong as the odor ave'sion, as measured by the control group, and interference with the taste aversion was shown by both exteroceptive and interoceptive cues (0212, CONTEXT-02). Recall that Experiment 3A, the high LiCl dose, was essentially a replication of Experiment 1, except in the general procedure of how animals were run through the chambers each day. An inspection of the figure reveals that the compound interoceptive and exteroceptive stimuli had opposite effects in these two experiments. In Experiment 3A, as in Experiment 1, both the aversions and the interference effects were larger for the odor component of the target. That is, the consumption measures for the control group (H20) looked the same as in Experiment 1, with the aversion for the odor cue considerably strange than that for the taste component. At the same time, the interference effects on 01 were larger. As in Experiment 1 there were no interference effects on the taste stimulus. These results indicate that the procedure change did not disrupt the basic potentiation effect or the relatively greater sensitivity of odor to disruption. Despite these consistencies, the major difference seen in these data is that the cue that interfered with avoidance of the target odor in Experiment 3A was the 0212 manipulation, as opposed to the compound exteroceptive group (SHOCK-02) in Experiment 1. In Experiment 1 only compound exteroceptive cues interfered with the aversion to the odor and the compound interoceptive cue (0212) did not. In Experiment 3A the compound interoceptive group interfered with conditioning while the compound exteroceptive cue did not. The important difference between these two experiments was the change in procedure MEAN ML CONSUMED (PLUS 0.1) MEAN ML CONSUMED 54 Almond test (target odor) Exp. 38 BASELINE WATER CONSUMPTION: 9.0 ML 6 _. N I- 4 " 8 cg - E c 0 8 2 _. o or I o — “9 "lntero- no bias "exteroceptlve" Interference ceptIve" Interference Condition Saccharlne Test (target taste) Exp. 38 BASELINE WATER CONSUMPTION: 9.0 ML 02 context no "lntefo- M b'“ ”exteroceptive" Interference ceptIve' Interference Condition 55 involving the delay room. Recall thatExperimentBB,thelowLiCl dose, wasa testofweakening the associative strength without changing the number of training trials. The results of Experimert 38 were similar to 3A, but as expected the lower dose of LiCl allowed interfererce effects not seer in Experiment 3A to emerge. Similar to Experiment 2 there were generally equal interference effects on the taste and odor components of the target. Both targets were attenuated by 0212, 02, & CONTEXT-02 in this expeiment (38). Also similar to Expeimert 2, the taste aversion shown by the control group (H20) was almost as strong as that of the odor aversion, which was not the case in either of the strong conditioning experiments (Experimerts 1 & 3A). Adjunctive behavior was not seer in Expeiment 3 and baseline water consumption was similar betweer the high and low dose conditions. Statistical analysis supported the above description. A 2 X 2 X 5 mixed ANOVA (dose by target by groups) was run on the data from Experiments 3A & 38. Dose and groups were the between subjects factor and target (01 81: T1) was the within subjects factor. This analysis revealed a significant main effect for groups, F(4,70) = 2.56, p < .05 (MSE = 10.9) and targets F(1,70) = 44.66, p < .01 (MSE = 205). The main effect for dose was not significant F < 1.0, nor were any of the interactions. Since the dose effect, and the interactions, were non significant, the data across doses were averaged and an Omnibus ANOVA was computed. This analysis revealed a significant main effect for target (01 vs T1), f(1,35) = 31.18, p < .01 (MSE = 1.92). However, no other significant effects were found: For groups, F(4,35) = 1.2, p > .05 (MSE = 10.28), and for the groups by target interaction, F(4,35) = 1.11, p > .05 (MSE = 1.92). Because the lack of interactions, produced by typically high variance, precluded post hoc tests based on these analyses one way AN OVA's were run on each of the target components in Experiments 3A and B. For the target odor (01) in Experiment 3A, F(4,35) = 2.51, p < .05 (MSE = 5.68). Duncan's Post hoc comparisons (p < .05 in all cases) were then conducted. The control group 56 (HZO) and CONTEXT-02 groups differed significantly from the 0212 group, while none of the other goups differed from each other. This confirms that in the high dose condition the only group that showed a significant difference from the control group was the compound interoceptive manipulation 0212. The O2 and Context groups, which were run as comparisons for the compound manipulations, showed consumption measures not reliably different from any other condition (for example the CONTEXT-02 condition), suggesting that these cues alone did not interfere with the target stimuli. If the compounds that included odor had shown reliably higher interference in all the experiments it would have suggested a summation effect, but the relative position of the compound versus the single cues reversed as after as it stayed the same, and little can be said regarding the contribution of odor to either exteroceptive or interoceptive interference manipulations. The one-way ANOVA for the target odor in Expeiment 3B yielded a significant effect, F(4,34) = 3.99, p < .05 (MSE = 4.57). The Post hoc analyses produced results similar to those of Experimert 3A, except that rather than the compound exteroceptive condition (CONTEXT-02) joining the control goup in differing from the 0212 goup (by lowering its consumption), it was the single stimulus exteroceptive stimulus (CONTEXT) that matched the H20 (control) goup in differing from the 0212 goup. As in the high dose condition, there were no other significant differences between goups. So in both the high and low dose conditions the only goup that differed significantly from the control goup was the compound interoceptive goup (0212). Another difference in the target odor tests was the substitution of single vs compound exteroceptive conditions in exhibiting reduced consumption, relative to the other experimental conditions, and thereby showing a non significant difference from the control goup. The one way AN OVA for the target odor in Experiment 3B yielded F(4,34) = 3.99, p < .05 (MSE = 4.57). The Post hoc analyses produced similar results to those of Experiment 3A except that rather than the compound 57 exteroceptive condition (CONTEXT-02) joining the control group in differing from the 0212 goup (by loweing its consumption) it was the single stimulus exteroceptive stimulus (CONTEXT) that matched the I-120 (control) goup in diffe'ing from the 0212 goup. As in the high dose condition, there were no other significant differences between goups. So in both the high and low dose conditions the only goup that differed sigrificantly from the control goup was the compound interoceptive goup (0212), and the major difference in the target odor tests was the substitution of single vs compound exteroceptive conditions that showed reduced consumption relative to the other goups and therefore did not differ from the control goup. The one way AN OVA for the taste target for Experiment 3A was not significant, F < 1.0. However, in the low dose condition, Experiment 3B, the ANOVA revealed significant results, F(4,35) = 4.04, p < .05 (MSE = 6.15). The Post hoc analyses show that the control goup differed from both the compound exteroceptive and interoceptive manipulations (CONTEXT-02 6: 0212) and, as in the odor test, it was the single stimulus exteroceptive cue (CONTEXT) that showed reduced consumption relative to the compound exteroceptive condition and it is the low consumption of this group that caused it to differ statistically from the the 0212 goup, p < .05 in all cases. Note also that the taste aversion shown by the control goup (HZO) was similar to the odor aversion, unlike the difference between target stimuli in the high dose control condition, Surprisingly, there was no obvious dose effect in our measures of the target stimuli. However, the dose effect became apparent on tests of consumption of the interference stimuli, as shown in the next data section. The manipulations that interfered among dose conditions also suggests differential dose effects, i.e. there was nonspecific interference in Experiment 3B, as in Expeiment 2, and more discrete effects in Expeiment 3A, as in Experiment 1. The mean consumption for the low dose condition (across goups and targets) was 2.2 ml while the mean consumption for the high dose 58 condition was 2.0 rrnl. An examination of baseline water consumption revealed no differences in initial consumption (baseline water consumption for high dose condition, in Box 1 = 9.1 ml, Box 2 = 8.9 ml; in the low dose condition, Box 1 = 9.0 ml, and Box 2 = 8.7 ml). Th in n ' uli The data for the interference stimuli are shown in Figures 9 and 10. A 2 X 5 ANOVA was computed for the interfering odor (02) data in Experiments 3A at B. The variables were goups and close. There was a significant main effect for goups, F(4,35) = 3.8, p < .01 (MSE = 19.25). Although the main effect for dose was not sigrificant, there was a sigrificant interaction, F(4,35) = 6.86, p < .01 (MSE = 6.58). Simple effects tests revealed significant differences between goups at both the low and high dose conditions. Duncan's Post hoc analyses showed that in Experiment 3A (high dose), all goups that encountered 02 in conditioning differed from those goups that did not encounter 02 during conditioning, except for the CONTEXT-02 goup. The control (HZO) and CONTEXT goups differed from each other, with the CONTEXT goup showing less neophobia. Post hoc analyses for Experiment 3B (low dose) showed that the H20 goup differed from all others and that the other goups did not differ from each other. Simple effects tests for each interference goup across the dose variable showed that three goups differed significantly irn their responses at the two different doses. The most important goup for the evaluation of the dose effect is the 0212 goup, which showed a difference in consumption between the dosages in the expected direction; that is, the aversion was attenuated by the reduction of the level of poisoning, F(1,35) = 4.80, p < .01 (MSE = 6.58) (mean rrnl consumed in Experiment 3A = 0.03, in 38 = 3.12). The control goup (1120) also consumed significantly more in the low dose condition, that is showed less neophobia, as would be expected at a lower dose of LiCl, MEAN ML CONSUMED MEAN ML CONSUMED 59 Banana test (Interference odor) Exp. 3A 8 I . BASELINE WATER CONSUMPTION: 3.9 M. a 4 q § ., 2 .1 $1 3 8 0 no ”Intero- "9 N“ ”exteroceptive” Interference ceptIve' Interference Condition 8 1 Banana test (Interfering odor) Exp SB BASELINE WATER CONSUMPTION: 3.7 ML 6 «n O 3 § 2 « g 2 o % A5232: ..... no ”Intero- no bias ”exteroceptlve" Interference ceptIve" Interference Condition MEAN ML CONSUMED 60 Qulnlne test (Interference taste) Exp. 3A BASELINE WATER CONSUWTION: 3.9 M. 02 context H20 02 ....... .f.:._' 3:. .... ................. ....... .............. ............ « -;-'.~:~' 02T2 .g.-..':_;‘.~.‘.;.. i-l-i‘.“ MEAN ML CONSUMED no ”Intero- "9 9‘“ ”exteroceptive" Interference ceptIve‘ Interference Condition Quinine test (interfering taste) Exp. SB . BASELINE WATER CONSUMPTION: 3.7ML 8 '._.;.' . 5.: j ......... W 02 context no "Intero- "9 bl“ ”exteroceptive” Interference ceptIve' Interference Condition 61 F(1,35) = 7.42, p < .01 (MSE = 6.58) (mean ml consumed in Expeiment 3A = 3.43, in 3B = 6.93). The CONTEXT goup also showed differential neophobia, but this time in the opposite direction, F(1,35) = 14.81, p < .01 (MSE = 6.58), mean ml consumed in Expeiment 3A = 7.75, and in 3B = 2.81. However this was a result of very high consumption in the presence of the novel banana odor in Experiment 3A relative to the other goups rather than a decrease in Experiment 3B. Recall that in Experiment 3A the CONTEXT goup also showed attenuated conditioning to 01. The 2 X 5 AN OVA (groups by dose) for the quinine interfeence taste showed no significant effects. That is, although consumption of quinine in Experiment 3B was slightly higher than in Experiment 3A, it was not significantly so, and in general the goups did not differ significantly among one another. Although a glance at the gaph suggests that only goups that had been conditioned to quinine showed lower consumption than those that were not conditioned to novel quinine, the differences were neither striking nor statistically sigtificant. The results for the inte'ference stimuli showed that all goups that had been exposed to these stimuli showed an aversion to them, and that in the low dose condition these aversions were slightly attenuated. This indicates that there were indeed associative effects to produce interference on the target and that the dose manipulation had a direct effect on the strength of these interfering effects Discussion m ’ nan n f rirnen 'hE 'men 1 2 To facilitate comparisons, Figures 11-13 show the results for the almond target, the saccharin target, and all eight target stimuli from the five goups common to all three experiments. Asterisks indicate the goups that showed significantly attenuated consumption measures relative to the control goup (HZO) within their appropriate experiment. It should be made clear that these composite gaphs are for the purposes of visual, not statistical, comparison 62 cozficoo 3:22.25 res—Eco ewe-.3353... rezieoeceier no... 2. .05-5.. II o H m m r a u 1 v . e .1‘ od wanngwg th(3 git—mg an an. :28 39.: ..e ESE? . . relieo 235.35.... BOD-aQOUOsO-IOI .88 0: 0050-87. 0‘ n W N a M m M \ .. m \ I M m w I n . . e a m ..e. t r o 1N0. "5% CE<~> 24mg r O u .axm cone :2... .8. SEE? :o_=o:oo 3:95:25 rezuneeecezer Iii-woo ac 11.3009 .2533 8.3.3.3... 0.39: o°~O~¢_s 96 Ian w u o u .1; —.a ”:98 ENE? 95% (a res—Eeeoceuae: es... 2. ..e-3c... O z .0. u u up .efl cone :9... so. 28E? rezooeo 2:35.52: f 1 i (l'O $an) cannsuoo 1w man I O .3 no.2 HzQ....t5mzoo ENE; 24mg . sum :23 .88: as 23E? To OBWDSNOO "I" NVSW :oEocoo 35.8.3... rezieo e”....e.e_.e.c. I.’—ufi.°°b.~l.l .I—‘ CC eOLCGC—l “.3000 txeruoo zo coEocoo 32.6.2.2:- res-Ken m .1 ad ”29% ENE; 53 3 sum .38. .85. .8» 8:388 63 rel-nee e..-5.3.3:. res_.eeoo.e.uer also: o..\\e\.c... e.. “.1003 2130 \\\\\\\\\\ txetuoo zo ! m S we. ”zQREszoo «Es; M.,-deem a dam .8»! 50.8. 32 5.2.85 3.3.3.3... seizes-9.3.3. es... 2. 5.3:... e.. o q . W o . N w W . a O m w m H r I. v o N Z ”w r v . m. .u r O 1 G v .5 ..e uzoznuzzmzoo 5:3 2.53 .l “ I (a den .32: :93. 88 5.2.88 e res-Eco 32:22.3... res..eeoe.e=er so... 2. 5.3:... o o r N r N 3 . w .e. an r v I r v . w m m 3 T O N I O u S 3.3 ioEuSwzoo «i=3 gamma .I o I o p 96 .8»: 30.8. 32 5.2.3.5 GSWI'ISNOO 'IW NVBW OBWI'ISNOO ‘IW NVBW MEAN ML CONSUMED MEAN ML CONSUMED MEAN ML CONSUMED MEAN ML CONSUMED (PLUS 0.1) N 64 Almond not (larger odor) Exp. 1 muramrmneesu. no 'telero- no Nee 'utueeeuln‘ Interference ...ll,.- Almond ted (target odor) Exp. 3A Wmmmrmuu 'Ielere- noel-o 'eneroeepuve' uprlve‘ Interference Condition no Interference Secchertn loot (forget taste) Exp I Tonnes Almond toot (target odor) Exp. 2 max WATER mm: 101 It ‘ _ i 3 1.1 no 'Innue- n- I‘u ‘oltoreelottvo' Interference u”...- . . Almond m1 (rug-t odor) EIP- 33 maximum-non 10“. can t 02 screen 2 a i I: i x I "‘ 'IeI-u. no Nee 'e-Iereeepttve' Interferon-e Interference Condition Seccnerln test (target teste) Exp 2 sun-rs WATER WM 102 u. . J 'lotereo "9 9'“ 'ellereeeoltve' no Interference “flu". Beechertn toot (tenet taste) Exp. 3A mummmmoiu no 'lotere- no the ‘elleroeepllve' Inlorlenneo eefllvo‘ Interference Condltlon no 'lnloro- Do..- 'utereeeolfro‘ Interference ..'.|'.- Beechertne Toot (teroet teste) Exp. 3! no 'tritoro- W. N" 'ureroeepllre' Inllrloronlo eeplleo‘ Interference Condltlon 65 and are provided in order to simplify the task of contrasting the results of the many conditions in these three expeiments. The question we set out to answer was whether thee are selective interference effects on odor and taste that result from exteroceptive and interoceptive interference events that occur between a target flavor and LiCl-induced illness. And although none of the interference effects were very large, the findings provide some support for selective inte'ference; similarly, there are a number of stable consistencies in the experiments that instill confidence. Generally, the results of Expeiment 3A (strong conditioning) were similar in many ways to those of Experiment 1 (strong conditioning) and the results of Expeiment 3B (weak conditioning) were generally quite like those of Expeiment 2 (weak conditioning). When the dose, and therefore the associative strength, of the situation was lowered, the interference effects wee more global on both the taste and odor targets and statistically significant interfeence was shown by the 0212 group on the odor target and by both the 0212 and CONTEXT-02 goups on the taste target. In the low dose condition the odor and taste aversions shown by the control goup were equivalent, while in the high dose condition the odor aversion exceeded the taste aversion. Another consistency among these experiments was the relative inability of any of the experimental manipulations to interfere with the 3233 aversions in the strong conditioning arrangements (using two trials or high LiCl). Furtlner, the interference effects tended to be stronger on the odor component of the target. However, thee were important differences in the pattern of interference between these expeiments. In Expeiment 1, only the compound exteroceptive goup (SHOCK-02) produced interference, although CONTEXT-02 followed close behind; but in Expeiment 3A only the compound in teroceptioe manipulation (0212) yielded interference. In fact, the inte'ference effects on the target odor in Experiments 1 and 3A were 66 reciprocals of each other. The only obvious difference between Experiments 1 and 3A was the backgound training procedure involving the delay room and the adjunctive behavior that occurred there. It is possible that procedures which promote adjunctive behavior have nonspecific stressful effects that engage an external defense strategy. For example, the procedure of Experiment 1 promoted high expectations of when water and food would be available, but access was blocked by restraint in the delay room. In contrast, Experiment 3 employed a procedure which minimized predictable waiting peiods, a condition known to produce adjunctive behavior (Falk, 1972), and it also produced little behavioral agitatiorn. While the indexing hypothesis assumes that the two defense systems operate relatively independently, a recent development suggests that there may be a reciprocal inhibitory relationship between them (Garcia, 1984, 1989; Garcia, Brett, 8: Rusirniak, 1989). That is, if one system is engaged, it tends to inhibit the activation of the other, as when the need to escape from a predator inhibits parasympathetic activation; and convesely, when exteroceptive activities, and sympathetic action, are inhibited after a large meal in favor of digestion and other visceral activities, like sleep. 1 It has been suggested that while the interoceptive system may be more "primitive," the exteroceptive system tends to take precedence in emergency situations in the same way as digestion is postponed when the sympathetic nervous system is activated (Suedfeld, 1980). In contrast to Experiments 1 G: 2, the conditions of Experiment 3 reduced significant environmental stressors. Under these circumstances, interoceptive interference was more pronounced and exteroceptive events had a much weaker impact. Thus, an interaction between general environmental stimulation and immediate proximal stimuli may have created differences in how the rat processed otherwise identical situations among these experiments. The compound exteroceptive goups showed variable effects throughout these experiments. Most notably, the effect of these manipulations changed dramatically between Experiment 1 and 3A. 67 This result supports the contention that the different procedures produced a shift in the general state of the subjects, altering the relative salience of the opposing classes of events, and thereby producing different patte'ns of interference. However, it is also fair to point out that thee were many inconsistent fluctuations in these data and that all these variations reflected a 3 or 4 rrnl attenuation of strong taste and odor aversions. Given the large number of treatrrnents, the likelihood of simple random fluctuation remains high. Another important trend, highlighted irn Expeiment 3, is that weaker conditioning parameters (e.g. Experiments 2 & 3B) promoted more general interference effects , with botln interoceptive and exteroceptive manipulations having more homogeneous effects on botln the taste and odor components of the target compound. And paradoxically, weak conditioning situations promoted stronger saccharin aversions than the strong conditioning expeiments. Note that consumption of T1 was reduced relative to the almond (01) consumption of the control goup (HZO) as compared to the high dose experiments, in which the odor aversion of the control goup was stronger than that of the taste avesion. This unexpected result, first seen in Expeiment 2, could be a result of "dual” conditioning to odor and "single" conditioning to taste. That is, animals presented with a taste-odor compound can avoid tasting the saccharin on the second training trial. In a single trial study this is not an issue. However, this result was also seen in Experiment 2, which had only one conditioning trial, suggesting that this may not be the case. Thus, some other explanation for the attenuating effect of taste potentiation of odor on the taste aversion itself may have to be entertained. Clearly, neither two conditioning trials nor high doses of LiCl are necessary to condition strong taste and taste potentiated odor aversions, although the LR rrnight be noted here that taste aversion learning does not appear to require consciousness, and may be formed during sleep. It has been reported that conditiorned food aversion occnrr even when the illness occurs under anesthesia (Bermudes-Rattoni, Forthman, Sanchez, Perez, & Garica, 1988; Roll 8: Srrnith, 1972). 68 relative strerngtln of odor to taste aversions in these two conditions are different. 5 I 1. . There are two different important comparisons in the present experiments: (a) the effects of different classes of inte'ference stimuli and (b) the effects of inte'ference on the taste aversion and its potentiated odor aversion. Before discussing these issues several things should be noted. In no case did the animals fail to learn about the target stimuli; that is, show drinking that approached baseline water consumption in the presence of the expeimental stimuli. Neither did subjects fail to learn about the interference stimuli interpolated between the target taste-odor compound and illness, that is, fail to reduce drinking in their presence. The aversions conditioned to both the taste and odor components of both the target and interference flavors were quite powerful and even the inte'ference effect was only a 4 rrnl attenuation. However, our control goups remained remarkably consistent and reliable across all the experiments, replicating previous reports of the potentiation effect; further our results were statistically significant despite the typically high variance seen in potentiation research. So although we were not able to observe massive differential interference effects we can have confidence that the results we did obtain were due to interference effects, and not with problems with the potentiation effect. The difficulty in disruptingtaste aversion learning is not unexpected. It has been repeatedly observed that leaning in this system is resistant to disruption (Bond, 1983; Der-Karabetian 6r Gorry, 1974; Kalat & Rozin,1971, 1972; Revusky,1971; Spear Gr Kucharski,1984 ); even brain damage may not render an organism incapable of taste aversion learning (Garcia, Hankins, & Rusirniak, 1974). Evidence for this comes from a series of brain lesion studies which were designed to determine whether the disruption of external and internal control systems could be neuroanatorrnically separated. While it was 69 possible to completely abolish shock avoidance learning with some lesions, such as of the amygdala, complete disruption of taste aversion learning was not accomplished. These studies irncluded a wide variety of brain areas krnown to be irnvolved in associative learning in the two systems. They included the ventral hippocampus and lateral septum, where lesions facilitated taste aversion learning but disrupted shock avoidance learrning; the amygdala, whee lesions that disturbed but did not abolish taste aversion learrning quashed shock avoidance leamirng altogether; the frontal cortex and olfactory bulb, where lesions inte'fered with both systems about equally depending on the area; the occipital cortex, where lesions disrupted shock avoidance learning more than taste aversion learning, but did not alter either system geatly; the dorsal hippocampus where manipulations facilitated both types of learning; and the medial septum, where lesions had a similar effect as those of the hippocampus but facilitated taste aversion learrning slightly more (Garcia, Hankins, Gr Rusirniak, 1974). It has been suggested that learning about interoceptive states is mediated by a very primitive brain system, and that involvement of the brainstem may protect its integity (Be'mudez-Rattoni et al., 1989; Grill, 1985;1(alat & Rozin, 1972; Norgen, 1974). In fact, complete brain transection at the level of the pons still leaves a decerebrate rat that can discriminate taste, demonstrate satiety, and acquire conditioned taste aversion (Norgen & Grill, 1982). So it comes as no geat surprise that the taste component of the taste-odor compound was difficult to disrupt. One question addressed in the present research was the extent to which the internal and external defense systems inte-fere with each other, and whether odor acquires the characteristics of taste through the potentiation process and is thereby insulated from inte'ference by other exteroceptive events. That odor was more easily disrupted was not unexpected. The crucial distinction was not whether odor was more sensitive to disruption than taste, but whether exteroceptive events disrupted the odor aversion more than the taste aversion since taste is known to be resistant to interference and odor is 70 thought to be primarily exteroceptive in nature. Our results show that the interfeence manipulation that interfered most often was the compound interoceptive (0212) one. This lends support to the indexing hypothesis and its predictions, since if the odor is incorporated into the visceral defense system, and the systems are somewhat independent, then interference should occur only when it comes from within the same system. However, given the size of the inte'ference effects, and the differential disruption shown when the expeimental parameters were changed, the present data hint at, but do not resolve the question of the protection of odor by conditionirng with taste or the functional insulation of the two defense systems. It is unfortunate that the temporal parameters for shock avoidance learrning are so short as to prohibit tlne interpolation of events between the stimulus and the shock reirnforcement, thus making it difficult to do complenentary experiments with shock. Interference effects were found to be more global in the weak conditionirng experiments. This was true whether the weak conditioning was done against a backgon of high arousal (Experiment 2) or occurred in a more "quiet" circumstance (Experiment 3B). Thus it cannot be concluded that this global interference was a result of inhibition of interoceptive learning by activation of the external defense system since in Experiment 3B the same effect was seen. Therefore, it can be assumed that in situations in which the conditioning is weak, because of either a lower LiCl dose or fewer conditioning trials, the aversion to the target flavor suffers when other salient events are present in the conditioning situation. That interference is more readily produced on the potentiated target supports the indexing hypothesis of potentiation. The within compound hypothesis would not predict such a difference. In fact, the within compound hypothesis implicitly predicts that odor should show just as much resistance to interference as taste because of its twofold activation in memory . Recall that according to this hypothesis odor and taste are associated with each other 71 and in addition both are associated with the US; during testing either stimulus is thought to activate both associative memories. An implication of this view is that the control exeted by the odor component of the compound should reflect the control exerted by the taste component (Westbrook et a1, 1983). Thus, the within compound hypothesis would predict equal interference effects in the present situation. Of course, a stimulus relevance, or preparedness dimension could be summoned to address this criticism, although the theory does not incorporate this variable at present. Given the above conclusions an important question is how strong is this protected, potentiated aversion relative to an odor aversion when it is conditioned alone? That is, what effect would the interference manipulations employed here have on an odor not indexed into the internal defense system by taste? This comparison would allow us to better appreciate the extent to which the taste brought the odor into the internal defense system and thus provided protection. If odor is primarily an exteroceptive cue that requires the presence of a taste to be conditioned into the visceral system, then the interjection of gustatory events during the CS-US delay should seiously attenuate any association of that odor with illness. The interpolation of anothe' exteroceptive event, on the other hand, should have little effect on the odors' association with illness since such an event is separate from botln the target odor and the illness. In this latter situation we might expect weak aversions by both exteroceptive events (the target odor and the interference stimulus). If, on the other hand, odor is not primarily an exteroceptive cue, and has gustatory biases even when presented alone, then additional gustatory experiences are not likely to interfere with its association with illness any more than they did in the present experiments, since we know from these data that rats easily learn about more than one gustatory event in this situation. Whatever the result, in addition to learning the characteistics odor possesses in this situation, we will have a measure of magnitude for odor conditioned alone with which to compare the present 72 data which will allow us to evaluate the effectiveness of taste in protecting it from interference. The present data have implications for current theories of associative learning, and for refining the details of the memorial indexing hypothesis. In its early formulations it was assumed that taste potentiated odor by protecting it from interference (Palme'ino et al., 1980). Later refinements of the model placed an emphasis on the "gating" or bridging phenomenon which suggested that this protection from interference reflected the incorporation of odor into a separate visce'al memory channel that has characteristics qualitatively different from those of the skin-defense system (Garcia & Rusirniak, 1980). Researchers were interested in whether taste aversions tolerate long delays because there are relatively few stimuli that can interfere with the memory of a gustatory event in this long-delay closed system or whether the memory for taste simply tolerates long delays in a general memory system as a result of a CS-US belongingness functionz- Taste aversions were found to be too robust for a clear test of this question. The present data do not directly address the issue of whether the systems are physiologically distinct, but the neurophysiological studies discussed above provide evidence that bear on that aspect of the model. The present data do suggest that taste incorporates odor into a system that is more vulnerable to disruption from other stimuli of the same type. That is, interference effects tend to be system dependent. A slow decay general systems model would not predict such results since if the memory for flavors Operates within the same system as do other event memories, but with different temporal parameters, there would be no reason to predict selective interference effects. While these data support the idea that the visceral system is more vulnerable to events of its own class, they also suggest that the exteroceptive and interoceptive systems interact in important ways that reflect the overall state of the organism. 2. There are, of course, different combinations of these premises that are entertained but this general description reflects the models that are the most influential. 73 REFERENCES Balsam, P. D. 8: Torrnie, A. (1985). Context and learning. Hillsdale N. 1.: Lawrence Erlbaum Associates. Bermudez-Rattoni, F., Forthman, D. L, Sanchez, M. A., Perez, I. L., 8: Garcia, (1988). Odor and taste aversions conditioned in anesthetized rats. flay}; ral Neuroscience _1_0_1_(5), 726-732. Be'mudez-Rattoni, F., Rusirniak, K. W., 8: Garcia (1983). Flavor-illness aversions: Potertiation of odor by taste is disrupted by application of novacaine into anygdala. WM 32, 61-75. Bermudez-Rattoni, F., Sanches, M. A., 8: Prado-Alcola, R. A. (1989). Learning of external and visceral cue consequences may be subserved by different neuroanatomical substrates. In T. Archer 8: L-G. Nilsson (Eds), W Hillsdale N.].: Lawrence Erlbaurn. Best, P. 1., Best, M. R., 8: Mickley, G. A. (1973). Conditioned aversion to distinct environmental stimuli resulting from gastrointestinal distress. lournal of Comparative and Physiological Psychology, _8_5_, 250-257. Bolles, R. C. (1979). learning Theory (Chap. 9, pp. 164-191). New York: Holt, Rinehart 8: Wilson. Bolles, R. C. (1970). Species-specific defense reactions and avoidance learning. Psychological Review, 22, 32-48. Bond, N. W. (1983). Reciprocal overshadowing in flavor aversion learning. my lournal of Em’ ental Psychology, 3§_B_, 265-274. Bouton, M. E., Jones, D. L., McPhillips, SA. 8: Swartzentruber, D. (1986). Potentiation and overshadowing in odor-aversion learning: Role of method of odor presentation, the distal-proximal cue distinction and the conditionability of odor. Learning and Motivation, 1_7, 115-138. 74 Bouton, M. E. 8: Whiting, M. R. (1982). Simultaneous odor-taste and taste-taste compounds in poison-avoidance learning. [gaming 8; mm IL 472-494- Breland, K. 8: Breland, M. (1961). The misbehavior of organisms. Amorican 3mm 11:. 681684. Capaldi, E. 1. (1967). A sequential hypothesis of instrumental learning. In K. W. Spence 8: 1. T. Spence (Eds), Tho Psychology of Learning and Mofiyagogflofl). New York: Academic Press. Coburn, K.L., Garcia, 1., Kiefe, S.W., 8: Rusiniak, KW. (1984). Taste potentiation of poisoned odor by temporal contiguity. Behavioral Neuroscience 28(5), 813-819. Der-Karabetian, A., 8: Gorry, T. (1974). Amount of different flavors consumed during the CS-UC interval in taste-avesion learning and interference. Wm 2(4), 457-460. Dickinson, A., 8: Mackintosh, N. 1. (1978). Classical conditioning in animals. Annual Review of Pagology, 29, 587-612. Domjan, M. (1983). Biological constraints on instrumental and classical conditioning: Implications for geneal process theory. Loaming and Motivation 1_7_, 215—277. Durlach, P. 1. 8: Rescorla, R. A. (1980). Potentiation rathe than overshadowing in flavor avesion learning: An analysis in terms of within-compound associations. WW Animal ghavior Hm, o(2), 175-187 Ellins, S. R., Thompson, L., 8: Swanson, W. E. (1983). Effects of novelty and familiarity on illness-induced aversions to food and place cues in coyotes (canis latrans) . lournal of Comparative Psychology, 27, 302-309. Ellins, S. R., 8: von Kluge, S. (1987). Preexposure and extinction effects of lithiun chloride induced taste-potentiated avesions for spatially contiguous auditory food cues in rats. Wong, flu), 164-169. 75 Ellins, S. R., 8: von Kluge, S. (In Press). Spatial contiguity not cue to consequence is the issue in taste-potentiated noise-illness associations: Reply to Holde et al. Behavioral Neurfl’ence Ellins, S. R., 8: von Kluge, S. (In Press). Auditory food cue conditioning: Effects of Spatial contiguity and taste quality. martely lournal of BMW Falk, 1. L. (1972). The nature and determinents of adjunctive behavior. In R. M. Gilbert 8:1. D. Keehn (Eds). MW Aggsion. Toronto: Addiction Research Foundation and University of Toronto Press. Galef, B. G. In, 8: Osborne, B. (1978). Novel taste facilitation of the association of visual cues with toxicosis in rats. lournal of Comparative and Phfiiologioal Psychology, 2;, 907-916. Garcia, 1. (1984). The evolution of learning mechamisms. Amorioan Psycholg‘ cal Association Master we Series: Psychology and gaming, Washington: American Psychological Association. Garcia, 1. (1989). Food for Tolman: Cognition and cathexis in concert. In T. Archer 8: L. G. Nilsson (Eds), Aversion, Avoidance, g; Anxioty. Hillsdale, N.1.: Lawrence Erlbaum. Garcia, 1., Brett, L. P., 8: Rusirniak, K. W. (1989). Limits of Darwinian Conditioning. In S. B. Klein 8: R. R. Mowre (Eds), Contemmrory Biological Constraints on Learning. Hillsdale N.1.: Lawrence Erlbaum. Garcia, 1., 8: Erwin, R. E. (1968). Appetites, Aversions, and Addictions: A Model for Visceral Memory. In 1. Wortis (Ed.), Recent Advances in W New York: Plenum Press. Garcia, 1. 8: Garcia y Robertson, R. (1985). Evolution of learning mechanisms. In B. L. Hammonds (Ed.), WM. Washington D. C.: American Psychological Association. 76 Garcia, 1., Hankins, W. G., 8: Rusiniak, K. W. (1974). Behavioral regulation of the milieu interne in man and rat. Eggnog, 185, 824-831. Garcia, 1., 8: Koelling, R. A. (1966). Relation of cue to consequence in avoidance learning. Wong, 4, 123-124. Garcia 1. 8: Rusiniak, K. W. (1980). What the nose learns from the mouth. In D. Muller-Schwarze 8: R. M. Silverstein (Eds), W N.Y.: Plenum Press. Garcia y Robetson, R. 8: Garcia, 1. (1985). X-rays and learned taste avesions: Historical and psychological ramifications. In T. G. Burish, S. M. Levy, 8: B. E. Meyewitz (Eds), W: A. W. Hillsdale N. 1.: Lawrence Erlbaum Associates. Grill, H. 1. (1985). Physiological mechanisms in conditioned taste avesions. Annals of the New York AW _4_4_3_, 67-88. Hankins, W. G., Garcia, 1., 8: Rusiniak, K W. (1974). Cortical lesions: Flavor illness and noise shock conditioning. Behavioral Biology, 1_Q, 173-181. Hankins W. G., Rusirniak, K.W., 8: Garcia, 1. (1976). Dissociation of odor and taste in baitshyness Behavioral Biology, _8_, 407-419. Hankins, W. G., Rusiniak, K. W., 8: Garcia, 1. (1976). Dissociation of odor and taste in shock avoidance learning. flovioral Biology, _1_8_, 345-358. Herrick, C. 1. (1948). The Brain of the Tigg Salamander. Chicago IL: University of Chicago Press. Herrick, C. 1. (1956). Evolution and Human Nature. Austin TX: Univesity of Texas Press. Holde, M. D., 8: Garcia, 1. (1987). Role of temporal orde and odor intensity in tastepotentiated odor avesions. Behavioral Neuroscience flu), 158-163. Holde, M. D., Leon, M., Yurmiya, R. 8: Garcia, 1. (1987). Effect of taste preexposure on taste and odor aveisons. Wand BehaviorI 15(1), 55-61. 77 Kalat, 1. W. 8: Rozin, P. (1971). Role of interfeence in taste-avesion learning. 0 -11-; 01 0 11 0 _-_ 1" ‘ a! 0. ' 1.. (<0 0m --._ ' a -1 019% 22(1), 53-58. Kalat,1. W. 8: Rozin, P. (1972). You can lead a rat to poison but you can't makehim think. InM. E. P. Seligman8:1. L. Hage (Eds),mal W- New York: Appleton-Century-Crofts. Kamin, L. 1. (1969). Predictability, surprise, attention, and conditioning. In B. A. Cambell 8: R. M. Church (Eds), Pong hmoo; ano Avgsivo mt, New York: Appleton-Century-Crofts Kaplan, P. 5., 8: Hearst, E. (1982). Bridging temporal gaps between CS and US in autoshaping: Insetion of othe stimuli before, during and after CS. lournal g m' ental Psychology: Animal Behavior Processes, _8_, 187-203. Kepple, G. (1982). Desigr_n and Analygs’ : A BogarM Handbook. Hillsdale N. 1.: Prentice-Hall, pp. 442-446. Kiefer, S.W. (1985). Neural mediation of conditioned food avesions. M d the New York Academy of Science. & 100-109. Kiefe, S.W., Rusiniak, K.W., 8: Garcia, 1. (1982). Flavor-illness avesion: Gustatory neocortex ablations disrupt taste but not taste-potentiated odor cues. lournal of @moarative and Phfiiological Psychology, 96(4), 540-548. Krane, R .V., 8: Robetson, D. P. (1982). Trace decay and the priming of short-tem memory in long-delay learning: Disruptive effects of novel exteroceptive stimulation. Learning and Motivation, _1_3, 434-453. Lett, B.T. (1973) Delayed reward learning: Disproof of the traditional theory. My; g and Motivation, 4, 237-246. Lett, B. T. (1974). Visual discrimination learning with a 1-min delay of reward. learning and Motivation, 5, 174-181. Lett, B. T. (1975). Long delay learning in the T maze. W Motivation. 6, 80-90. 78 Let, B. T. (1984). Extinction of taste aversion does not eliminate taste potentiation of odor avesion in rats or color avesion in pigeons. Animal Learning and Behavior, 1_2, 414-420. Logue, A. W. (1979). Taste avesion and the geneality of the laws of learning. Psygological Bullgg’ , &(2), 276-296. LoLordo, , M. 8: Droungas, A. (1989). In S. B. Klein 8: R R. Mowre (Eds), Contemmmy Learning Theories: Instrumental Conditiong Theory WWW Hillsdale N. I -= Lawrence Erlbaum. Mackintosh, N. 1. (1974). The Psygology of Animal Learning. New York: Academic Press. Mackintosh, N. 1. (1975). A theory of attention: Variations in the associability of stimuli with reinforcement. Psychological Review 82(4), 276-298. McGowan, B. K. Hankins, W. G. 8: Garcia, 1. (1972). Limbic lesions and control of the internal and external environment. Behavioral Biology Z, 841-852. Mikulka, P. 1., Pitts, E., 8: Philput, C. (1982). Overshadowing not potentiation in taste aversion conditioning. Bulletin of the Psychonomic Society, 29(2), 101-104. Nachman, M., 8: Ashe, 1. L. (1973). Learned taste aversion in rats as a function of dosage, concentration, and route of administration. Physiology and M4 .19. 753-755- Norgren, R. (1974). Gustatory affeents to ventral forebrain, Brain Research 81, 285-295. Norgren, R., 8: Grill, H. (1982). Brain-stem control of ingestive behavior. In D. W. Pfaff(Ed.), Tho Physiological mechanisms of motivation. N. Y.: Springer-Verlag. Palmerino, C., Rusirniak, K. W., 8: Garcia, 1. (1980). Flavor-illness aversions: The peculiar roles of odor and taste in memory for poison. Science, 2%, 753-755. 79 Pavlov, I. P. (1927). Conditioned reflexee. London: Oxford Univesity Press. Petrinovich, L., 8: Bolles, R. (1957). Delayed alternation: Evidence for symbolic processes in the rat, lournal of Qmmative and Physiological Psychology. 59, 363-365. Rescorla, R. A. (1988). Pavlovian conditioning, it's not what you think it is, W 433, 151-160. Rescorla, R. A., 8: Cunningham, C. L. (1978). Within compound flavor associations, 0 --. 01 41-11.- 11 =11 [3. h -1 0101a . ‘ L. o - L‘ grow o, 207-216. Rescorla, R. A. 8: Durlach, P. 1. (1981). Witlnin-event learning and Pavlovian conditioning. In N. E. Spear 8: R. R. Miller (Eds), Information WW Hillsdale N.1.: Lawrence Erlbaum Associates. Rescorla, R. A. 8: Wagne, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black 8: W. F. Prokasy (Eds), elm Conditioning 11: Current ResearQ and Ther. New York: Academic Press. Revusky, S. (1971). The role of interfeence in association ove a delay. In W. K. Honig 8: P. H. R. 1ames (Eds), Animal Memogy. New York: Academic Press. Revusky, S. (1971). The Concurrent interference approach to delay learning In M. 1. Barker, M. R. Best, 8: M. Domjan (Eds), learning Mechflisms 131mm Waco TX: Baylor Univesity Press. Revusky, S., 8: Garcia, 1. (1970). learned associations ove long delays. In G. H. Bower (Ed.), The Poygzology of learning and Motivation: Advances in Reserach and Theogy. New York: Academic Press. Roll, D. L. 8: Smith (1972). Conditioned taste avesion in anesthetized rats, In M. E. P. Seligman 8: 1. L. Hage (Eds), Biological Bound_r_ies of Learning. New York: Appleton-Century-Crofts, pp. 98-102. 80 Rozin, P. (1976). The evolution of intelligence and access to the congitive unconscious. WWW goo. New York: Academic Press. pp. 245-281. Rozin, P., 8: Kalat, 1. W. (1971). Specific hunges and poison avoidance as adaptive specializations of learning. Mologig Review, _7_§, 459-486. Rozin, P., 8: Schull, 1. (1989). The adaptive-evolutionary point of view in experimental Psychology. A critical analysis of "constraints". In R. C. Atkinson, R. 1. Hemstein, G. Lindzey, 8: R. D. Luce (Eds), M Handbook of Emrimeetal Peyghology. New York: Wiley. Rudy 1. W., Rosenberg, L., 8: Sandel, 1. H. (1977). Disruption of a taste familiarity effect by novel exteroceptive stimulation. lournal of WW 3(1), 26-36- Rusirniak, K.W., Garcia, 1.8: Hankins, W.G. (1976). Bait-shyness: Avoidance of the taste without escape from the illness in rats. lourng of Comparative and Phfiiological PeyQolgy, 29(5), 460-467. Rusiniak, K. W., Hankins, W. G., Garcia, 1., 8: Brett, L. P. (1979). Flavor-illness avesions: Potentiation of odor by taste in rats. Behavioral e Neural Bmgy, 25, 1-17. Rusiniak, K. W., Palmerino, C., Rice, A. G., Forthman, D. L., 8: Garcia, 1. (1982). Flavor-illness aversions: Potentiation of odor by taste with toxin but not shock in rats lournal of Comparative and Physiological Psymoiogy, %, 527-539. Seligman, M. E. P. (1970). On the geneality of the laws of learning. 25mm zz 512-53. Sjoden, P-O., 8: Archer, A. (1989). Taste-aversion conditioning: The role of contextual stimuli. In T. Archer 8: L-G. Nilsson (Eds), Aversion Avoidancel and Anxiety. Hillsdale N. 1.: Lawrence Erlbaum Associates. 81 Spear, N. E., 8: Kucharski, D. (1984). Ontogenetic diffeences in the processing of multi-element stimuli. In: H. L. Roitblat, T. G. Beve, 8: H. 5. Terrace (Eds), Animal Cogg’tion. Hilldale N. 1.: Lawrence Erlbaum. Staddon, 1. E. R., 8: Simmelhag, V. (1971). The "superstition" experiment: A reexamination of its implications for the principles of adaptive behavior. W 18(1),16-43. Suedfeld, P. (1980). Restricted Environmeiitfitimuletion: Research and Applioatione. N.Y.: Wiley. Weine, B. 1. (1962). Statistical Principles in m' ental Desigg. New York: McGraw Hill, pp. 529-532. Westbrook, R. P., Clarke, 1. C. 8: Provost, S. (1980). Long-delay learning in the pigeon: Flavour, color, and flavour-mediated color avesions. Behavoral and Neural Biology, 284 398-407. Westbrook, R. F., Homewood, R. P., Horn, K., 8: Clarke, 1. C. (1983). Flavour-odour compound conditioning: Odor-potentiation and flavour attenuation. The Qper_tely lournal Q Qperimental Psychology. 3_5_l_3, 13-33. "Illlllllllllllllllll