Operational terminology for stimulus exposure (SE) conditioning

Behavioural Brain Research 95 (1998) 143 – 150

Review article

Operational terminology for stimulus exposure (SE) conditioning Glen D. Brown * Computational Neurobiology Laboratory, The Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, CA 920037, USA Received 5 January 1998; received in revised form 27 April 1998; accepted 27 April 1998

Abstract Nomenclature for simple types of learning is ambiguous and incomplete, as even commonly used terms such as ‘habituation’ and ‘sensitization’ are not applied consistently. One problem is a failure to distinguish between operational and theoretical constructs linguistically. Operational terminology for reporting behavioural results should be different from the language used in the discussion of learning theory. Thus, systematic operational terminology for simple types of conditioning is proposed. The most general category, stimulus exposure (SE) conditioning, is learning by exposure to a stimulus or to multiple stimuli where explicit inter-stimulus contingencies or instrumental reinforcers are not a part of training. Subtypes of SE conditioning are distinguished by the number of different stimuli used during training, by the method used to assess learning, and by the relationship between training stimuli and the assessment method. These categories include ‘alteractive’, ‘iterative’, ‘heterostimic’, and ‘multistimic’ conditioning. Learned responses are also categorized as reduction, enhancement, or transformation. SE conditioning categories combine with response terminology in phrases such as ‘iterative reduction’, which is a decrease in the response to a stimulus due to repeated presentation of that stimulus. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Habituation; Inhibition; Learning; Memory; Nonassociative; Sensitization

1. Introduction The importance of distinguishing operational and theoretical constructs in the study of learning is widely recognized [22,26,29,40]. Operational constructs are observational, based on types of experience or characteristics of a learned response. In contrast, theoretical constructs are hypothetical nervous system processes, and they can take different forms depending on the theoretical context. Theoretical learning processes underlie operationally named learned behavior, and therefore, learning processes can sometimes be inferred by observing learned behaviour. Unfortunately, the two levels of construction are sometimes confused.

* Tel.: +1 619 4534100 ext. 1124; fax: + 1 619 5870417; e-mail: [email protected]

The causal relationship between learning processes and behaviour makes it more rather than less important to use unambiguous terminology to distinguish the communication of experimental results from the exposition of learning theory. Unfortunately, the learning and memory vernacular does not always support this distinction. Many terms have been used in both an operational sense and a theoretical sense. For example, the term ‘habituation’ [14,18,19,41] has been defined as a response decrement that occurs due to repeated presentation of a stimulus (operational) and as a learning process that filters out repetitive and therefore innocuous stimuli (theoretical). Thompson et al. [40] recommended reserving the term ‘habituation’ for the theoretical learning process, and suggested the phrase ‘response habituation’ as an operational specifier. However, this leads to the confus-

0166-4328/98/$ - see front matter © 1998 Elsevier Science B.V. All rights reserved. PII S0166-4328(98)00071-0

144

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

Fig. 1. Assessment methods used in conditioning experiments. Pulse marks represent discrete testing stimuli. Plaid bar represents discreet training stimuli. Learning is demonstrated by comparing trained animals to untrained (nave) controls with a post-training test stimulus (A). It is also possible to assess learning by comparing the response to a stimulus before (pretest) and after (posttest) training (B). However, at least one additional comparison is necessary in this case—between pre-test and post-test in a control group not getting the intervening training. During assessment in either case, spontaneous behaviour can be measured instead of the response to a test stimulus.

ing possibility that response habituation is produced by something other than habituation since there are multiple distinguishable learning processes that can cause a response decrement [2,36]. Peeke and Petrinovich [29] agreed that the term ‘habituation’ should be reserved for the theoretical learning process and also noted problems associated with the phrase ‘response habituation’. They suggested that learned behaviours be named according to ‘observables such as response decrement, increase in latency, or reorganization of the pattern of behavior’. This advice has not always been heeded however, and the term ‘habituation’ is still used routinely as an operational specifier [4]. This is because an observation such as ‘a response decrement due to repeated presentation of the same stimulus’ is more conveniently communicated with a single word or short phrase. The term ‘sensitization’ also has a number of operational and theoretical definitions. Perhaps the most common operational usage is the facilitation of a response to one stimulus due to prior exposure to an alternate stimulus (discussed in [8]). As a learning process, ‘sensitization’ is an experience-dependent, generalized excitation similar in many ways to arousal [12,19]. In order to help separate operational and theoretical levels of analysis, systematic operational terminology for simple types of learning is proposed here. The evolutionary significance of learning processes is also emphasized [23,42], although this particular theoretical framework is not critical to the proposal. The terms ‘habituation’ and ‘sensitization’ are used only in their theoretical sense.

2. SE conditioning In conditioning experiments, learning is assessed by comparing a group of animals that has received training (experimentals) to a group of animals that has not (naive controls; Fig. 1A). Another possibility is to compare the same animals before and after training (but see [35]). To control for the assessment procedure as well as other factors such as development, a control group that does not receive training is included for comparison in this case (Fig. 1B) (see [30] for another approach). After a behavioural change is detected by one of these methods it can be named operationally. Of the variables present in a conditioning experiment (Table 1), only the training procedure and the assessment method can be easily and systematically manipulated by investigators. Discrete, reproducible stimuli are often an important part of both training and assessment in the laboratory [35]. Kimble [22] referred to the stimulus paradigm as the independent variable in conditioning experiments. Thus, the nomenclature proposed here is based only on training and assessment procedures. Rescorla [35] described three types of experience that can be studied in conditioning experiments: (1) simple exposure to stimuli; (2) interstimulus contingencies; and (3) instrumental reinforcement. The operational terminology proposed here is concerned only with the first of these. The phrase ‘stimulus exposure (SE) conditioning’ will refer to all beha6ioural changes produced by exposure to

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

stimuli where neither instrumental reinforcement nor interstimulus contingencies are not an explicit part of training. A single stimulus can be presented randomly or in a pattern since intrastimulus contingencies are not excluded by this definition (e.g. [6]). It is difficult to identify and eliminate every interstimulus contingency or instrumental reinforcer present in an SE conditioning experiment. Similar problems arise in classical and instrumental conditioning research [5,26]. Nevertheless, it remains useful in practice to study learning where a well-controlled training stimulus is presented at random times both with respect to other types of stimuli and with respect to behaviour [13]. The phrase ‘nonassociative learning’ has also been used to group simple types of conditioning [19]. The concept of associativity in learning theory arises from the observation that learning often depends on a contingency present during training. Examples include Pavlovian learning and instrumental learning, which are thought to be the result of selective pressures to learn about contingencies between stimuli or between a stimulus and a behaviour, respectively. In contrast, nonassociative learning processes do not depend on contingencies present during training. However, associations can be made in even the simplest types of conditioning experiments. For example, context specificity, the expression of a learned behaviour only in the environment where training has occurred, may be the result of associations made between a stimulus and the training environment [27,32,37,44]. Wickens and Wickens [45] studied other apparently associative processes induced by simple exposure to stimuli. In general, it takes more controls than those shown in Fig. 1 to determine if learning is associative or nonassociative. Thus, while the associative-nonassociative distinction will be retained to discuss theoretical learnTable 1 Variables in conditioning experiments Training stimulus paradigm properties: Number of different stimuli Number of trials with each stimulus Fixed stimulus properties (intensity, duration, modality) Interstimulus contingencies Intrastimulus contingencies Instrumental reinforcement Assessment method Presence or absence of an alpha response Modifiable stimulus properties (neutral, aversive, appetitive, etc.) Learned response properties: Intensity, duration, qualitative change, etc. Persistence Internal state (motivation, illness, etc.) Physiological changes Underlying learning processes

145

ing processes [2], the concept of associativity cannot be used in an operational nomenclature.

3. Terminology The proposed SE conditioning nomenclature is based on: (1) the number of different stimuli that are used during training; (2) whether or not a discreet stimulus is used to assess learning; and (3) if a stimulus is used to assess learning, whether or not it was also used during training (Figs. 2 and 3). Implicit in the following definitions is that training stimuli are not reinforcing any behaviour. ‘Alteractive conditioning’ is any change in spontaneous beha6iour due to the presentation of one type of stimulus during training. ‘Iterative conditioning’ is any change in the response to a test stimulus due to training with that same stimulus. ‘Heterostimic conditioning’ is any change in the response to a test stimulus due to training with an alternate stimulus. Alteractive, iterative, and heterostimic conditioning are produced by exposure to a single type of stimulus during training, and together are called ‘monostimic conditioning’. ‘Multistimic conditioning’ is any change in beha6iour due to training with more than one type of stimulus where no interstimulus contingencies exist during training. Multistimic conditioning can be subdivided into type A if a spontaneous behaviour is used to assess learning, type B if the test stimulus was also used during training, or type C if the test stimulus was not used during training (Fig. 2). If necessary, other characteristics of the stimulus paradigm may be specified in the usual way. Differences in the number of training trials used to produce SE conditioning may be distinguished by prefacing the learning category with phrases such as ‘single-trial’ or ‘multiple-trial’. An experiment may be described as ‘fixed interval’ or ‘variable interval’ depending on the nature of the intertrial interval (ITI). An operational definition for ‘instrumental conditioning’ can be given in this context as any change in beha6iour due to training with one or more stimuli used to reinforce beha6iour [38,43]. Similarly, ‘classical conditioning’ is any change in beha6iour due to training with more than one type of stimulus where an interstimulus contingency exists during training [28,34]. These are the usual operational definitions for classical and instrumental conditioning [26].

4. Learned responses When an observable response to a stimulus exists in

146

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

Fig. 2. Conditioning paradigms. S1, S2, and S3 are discrete stimuli presented at pulse marks. Flat lines (B1) represent the measurement of a spontaneous behaviour. Alteractive and type A multistimic conditioning are changes in a spontaneous behaviour. Iterative and type B multistimic conditioning are changes in a response to a stimulus that is also used during training. Heterostimic and type C multistimic conditioning are changes in a response to a stimulus that is not used during training. Classical (type B) and instrumental (type A) conditioning paradigms are shown for comparison.

naive subjects, it is called an a-response [17,21]. The following conventions were adopted: (1) learning characterized by the strengthening of an a-response or an increase in a spontaneous behaviour was called ‘enhancement’; (2) the waning of an a-response or a decreased spontaneous behaviour was called ‘reduction’ or sometimes ‘decrement’; and (3) a qualitative change in a response was called a ‘transformation’. The SE conditioning categories defined above can be combined with response terminology to form operational names in SE conditioning experiments. An increase in a response due to repeated presentation of a stimulus is iterative enhancement (e.g. [7]; used in [4]). An increase in a spontaneous behaviour due to exposure to one type of stimulus is alteractive enhancement (e.g. [39]). A qualitative change in the response to one stimulus due to exposure to another stimulus is a heterostimic transformation (e.g. [9]).

As mentioned, learned behaviour should be described as accurately as possible [29], but succinct operational classifiers are also necessary. For example, the phrase ‘reduction in the probability of responding to a chemical stimulus due to prior experience with a mechanical stimulus at an alternate site’ is more informative than ‘heterostimic reduction,’ but the latter is a useful shorthand (e.g. [2]). Enhancement, reduction, transformation, the establishment of a response to a stimulus that previously elicited no detectable response, or any other change in behaviour that occurs as a result of training may also simply be called ‘conditioning’. Other descriptors of the learned response can be added in the usual way. For example, the duration of a response change may be noted in phrases such as ‘long-term multistimic conditioning’ [2].

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

147

Fig. 3. Flow chart for reporting results of stimulus exposure (SE) learning experiments. The subtypes of multistimic conditioning, A, B and C, can be differentiated by the last two questions which differentiate alteractive, iterative and heterostimic, respectively.

5. Example: iterative conditioning In an iterative conditioning experiment, a stimulus is presented for one or more trials during training and then is used to test for learning (Fig. 2). One or more learning processes must underlie iterative conditioning, and one candidate is habituation. Thompson and Spencer [41] identified many of the common behavioural manifestations of habituation. The most prominent of these is a decrease in the a-response (a more complete list may be found in [2]). However, when inferring a learning process from behavioural observations, absolutes rarely obtain. Consider three potential difficulties in concluding that habituation is the learning process under study in iterative conditioning experiments. First, habituation can effect different response components in different ways. Multiple changes in behaviour often occur in one experiment, and the latency, duration, and probability of a response do not necessary change in parallel [1,4,16,31]. Similarly, habituation may cause a decrease in one response that unmasks a competing response. (In the past, there has been no way to refer to an entire set of response changes in an

experiment of this sort whether or not they are a result of habituation. With the proposed terminology, the entire set or any subset of response changes can be classified as iterative conditioning.) Second, other learning processes besides habituation may be manifest in iterative conditioning experiments. The iterative reduction in one response component may be the result of habituation while the reduction in another is not [2,36]. Excitatory learning processes such as sensitization also shape responses in iterative conditioning experiments [12]. Finally, if no a-response is present, then it is not possible to observe a reduction as defined above. Habituation can still be inferred however, for example by the failure of an iterated stimulus to form associations [24,33]. (Whether or not latent inhibition and the USpreexposure effect are related to habituation is beyond the scope of this work). Another theoretical learning process that produces iterative reduction is learned taste aversion (LTA) [10]. Animals do not approach and eat a poisoned food stimulus that has previously elicited a consummatory response. It should not be disconcerting that learning processes as disparate as habituation and LTA underlie

148

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

Table 2 Converting to systematic terminology Former name(s)

Defined

Systematic name

Habituation (adaptation)

Response decrement due to repeated presentation of the same stimulus

Iterative reduction

Facilitation (sensitization, windup)

Response increment due to iterative training

Iterative enhancement

Sensitization (potentiation, pseudo-conditioning)

Response increment due to presentation of an alternate stimulus

Heterostimic enhancement

Inhibition (negative sensitization)

Response decrement due to presentation of an alternate stimulus

Heterostimic reduction

Sensitization

Nonassociative facilitation of an a-response

Enhancement*

Pseudo-conditioning

Nonassociative novel response

Transformation*

* Use of these terms does not imply that learning is nonassociative.

the same operational type of learning. Operational constructs are not meant to be the subject of comparative study. Theoretical processes are the appropriate topic for the comparative study of learning; operational terminology is simply the language of comparative work. In general, one learning process can underlie multiple operational types of conditioning, and multiple learning processes can produce the same operational type of conditioning (see [26] for a similar discussion on classical and instrumental conditioning).

6. Other operational constructs ‘Iterative reduction’ and ‘heterostimic enhancement’ replace the operational uses of ‘habituation’ and ‘sensitization’, respectively, discussed above (Table 2; [3]). Kandel [20] proposed the phrase ‘negative sensitization’ to describe a decrease in the response to one stimulus due to prior presentation of an alternate stimulus, but the phrase has not been adopted [25]. ‘Negative sensitization’ in its operational sense is synonymous with the new phrase ‘heterostimic reduction’ (Table 2). The term ‘inhibition’ has also been used as a synonym for heterostimic reduction ([25]; Table 2). ‘Inhibition’ (or ‘conditioned inhibition’) already has at least one theoretical usage in classical conditioning research [26,28,34,35]. Learned excitation is the learning of a positive correlation between stimuli. Inhibition is learning that opposes learned excitation, for example by the learning of a negative correlation. Unlike the definitions proposed here, inhibition is not defined by its effect on the a-response. Learned excitation can diminish an a-response while inhibition can enhance it. Thus, ‘reduction’ is used here instead of ‘inhibition’ when learned behavior is to be described operationally. The term ‘inhibition’ is retained to describe a nonassociative learning process [2,25], although its relationship to conditioned inhibition remains unclear.

The enhancement of a response due to repeated exposure to one type of stimulus has been called ‘sensitization,’ ‘early facilitation’, ‘use-dependent facilitation’, and ‘windup’ (e.g. [7]). The latter three choices are strictly operational but do not fit into a larger scheme such as the one proposed here. The new name for this type of learning is ‘iterative enhancement’ (Table 2; [4]). One operational definition for ‘pseudoconditioning’ is the establishment of a new response or the qualitative change in an existing response due to the simple presentation of a stimulus, though the term has been used in other contexts as well [9,11,21,22]. Qualitative changes in a response are called ‘transformations’ in the present work (Table 2). The term ‘pseudoconditioning’ may be removed from the lexicon altogether unless it can be distinguished from sensitization as a learning process. Iterative, heterostimic, and multistimic conditioning categories are similar to conditioned stimulus (CS)only, unconditioned stimulus (US)-only, and random controls, respectively, in classical conditioning research. However, the terms ‘CS’ and ‘US’ are not useful outside of a classical conditioning context. Similarly, an alteractive control is often used in instrumental conditioning experiments. Assessment procedures in conditioning experiments can be more complicated than Figs. 1–3 imply [24,33]. Rescorla [35] divided learned changes in behaviour into three categories, including changes in response evocation, in learnability, and in value. Only changes in response evocation have been considered so far. Learning assessed by other methods could also be named with the proposed SE conditioning terminology if desired, but the practical usefulness of doing so is unclear. The proposed nomenclature is not meant to add another name to operational constructs that are already unambiguously defined in classical or instrumental conditioning research. More generally, it is not necessarily advantageous to assign an operational name to every learned behaviour

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

examined in the context of a larger study. For example, if the stimulus specificity of a learned response is tested by changing the type of conditioning from iterative (or type B multistimic) to heterostimic (or type C multistimic), then the established concept of stimulus specificity can be retained to describe the results (e.g. [2]).

7. Theoretical constructs and neural implementations The SE conditioning stimulus paradigm used to produce and assess learning does not necessarily imply anything about underlying learning processes. For example, the distinction between iterative and heterostimic conditioning is based on whether or not, respectively, the training stimulus is used to test for learning (Figs. 2 and 3). By design however, learning only takes place during training (Fig. 1). Because the same training regime is used, the only possible difference between iterative and heterostimic conditioning is performance at the time of testing, not whatever learning occurred during training. Different learning processes may or may not be manifest depending on the assessment procedure [35]. Even when operational categories are based on the stimuli presented during training (e.g. monostimic, multistimic, classical, and instrumental conditioning, Fig. 4), these categories will not necessarily distinguish the underlying learning processes [5,26]. Categorizing SE conditioning only requires that a learned change in behaviour be evident at the time of testing (Fig. 1). In general, classifying underlying learning processes requires more information such as results of control experiments or prior knowledge about the learned behaviour under study. For example, establishing that Pavlovian learning is present may be accomplished with control procedures more elaborate than those described in Fig. 1. Classically conditioned subjects are compared to a multistimic control group in which the interstimulus contingency of interest is eliminated [34]. If learning does not occur in the multistimic group, then Pavlovian learning is more likely to be the process mediating the change in behaviour observed in the classically conditioned group. Similarly, the parametric features of habituation [41] may be thought of as control procedures for demonstrating that habituation is present. Iterative reduction that undergoes spontaneous recovery, that occurs more rapidly in repeated training sessions, and that can be reversed by an alternate stimulus is probably due to habituation [2]. The ethological view is that learning allows animals to use experience to choose behaviours that increase

149

their fitness [23,42]. Similar learning processes exist in different species or across sensory modalities either because of convergent evolution or common heritage. Thus, these learning processes can be grouped by their hypothesized adaptive value. For example, it would seem to be adaptive for animals to learn when food has made them sick, and LTA appears to be a ubiquitous form of learning [10]. Another class of theoretical construct identified by Kimble [22] was hypotheses about the neural substrates of learning. Rather than a type of theoretical construct however, neural mechanisms can be considered a separate level of study. In this formulation, theoretical constructs are abstracted away from hypotheses about neural mechanisms. For example, habituation is a learning process that allows animals to screen out repetitive or unreinforced stimuli (theoretical). One behavioural manifestation of habituation is iterative reduction (operational). A candidate neural mechanism for habituation is use-dependent depression at sensoryto-motor and sensory-to-interneuron synapses [15]. Many of the same issues concerning theoretical and operational constructs also apply to the study of neural substrates. For example, a learning process such as habituation can in principle have more than one neural implementation. The characterization of theoretical learning processes and the study of their neural substrates are important goals in neuroscience. It is hoped that a standardized language for reporting behavioural results from SE conditioning studies will contribute to this effort. Implementing the terminology proposed in the present work should occur in stages. First, researchers must begin to distinguish between operational and theoretical levels of analysis on a consistent basis. Second, names for operational types of learning that do not have commonly used alternatives (e.g. iterative enhancement and multistimic conditioning) may be adopted. Finally, if the terminology does indeed offer practical advantages, investigators should replace operational usage of theoretical constructs (e.g. habituation or sensitization) with the systematic alternatives (Table 2).

Acknowledgements Thanks to Drs Robert Podolsky, Alan Randich, Terrence Sejnowski, and Satoru Shiono for helpful comments on earlier versions of the manuscript, and to Friday Harbor Laboratories for providing an inspirational working environment. A National Institute of Mental Health predoctoral fellowship (5F31 MH10006) and a National Science Foundation postdoctoral fellowship (INT 9311719) supported this work.

150

G.D. Brown / Beha6ioural Brain Research 95 (1998) 143–150

References [1] Barry RJ, Sokolov EN. Habituation of phasic and tonic components of the orienting reflex. Int. J. Psychophysiol. 1993;15:39 – 42. [2] Brown GD. Nonassociative learning processes affecting swimming probability in the seaslug Tritonia diomedea: habituation, sensitization, and inhibition. Behav. Brain. Res. 1998;95:151 – 65. [3] Brown GD. Heterostimic enhancement of a not so fixed action pattern. Neth. J. Zoo. 1994;44:184–99. [4] Brown GD, Frost WN, Getting PA. Habituation and iterative enhancement of multiple components of the Tritonia swim response. Behav. Neurosci. 1996;110:478–85. [5] Brown PL, Jenkins HM. Auto-shaping of the pigeon key-peck. J. Exp. Anal. Behav. 1968;11:1–8. [6] Davis M. Effects of interstimulus interval length and variability on startle-response habituation in the rat. J. Comp. Physiol. Psychol. 1970;72:177–92. [7] Ehrlich JS, Boulis NM, Karrer T, Sahley CL. Differential effects of serotonin depletion on sensitization and dishabituation in the leech Hirudo medicinalis. J. Neurobiol. 1992;23:270–9. [8] Eisenstein EM, Eisenstein D, Bonheim P. Initial habituation or sensitization of the GSR depends on magnitude of first response. Physiol. Behav. 1991;49:211–5. [9] Erickson MT, Walters ET. Differential expression of pseudoconditioning and sensitization by siphon responses in Aplysia: novel response selection after training. J. Neurosci. 1988;8:3000 – 10. [10] Garcia J, Hankins WG, Rusiniak KW. Behavioral regulation of the milieu interne in man and rat. Science 1974;185:824– 31. [11] Grether WF. Pseudo-conditioning without paired stimulation encountered in attempted backward conditioning. Comp. Psychol. 1938;25:91 – 6. [12] Groves PM, Thompson RF. Habituation: a dual-process theory. Psychol. Rev. 1970;77:419–50. [13] Hall G. Exposure learning in animals. Psychol. Bull. 1980;88:535 – 50. [14] Harris DJ. Habituatory response decrement in the intact organism. Psychol. Bull. 1943;40:385–422. [15] Hawkins RD, Clark GA, Kandel ER. Cell biological studies of learning in simple vertebrate and invertebrate systems. In: Mountcastle VB, Plum F, Geiger SR, editors. Handbook of Physiology, Section 1: The Nervous System. vol. 5. Higher Functions of the Brain, Part I. Bethesda: American Physiological Society, 1987:25 – 83. [16] Hinde RA. Behavioral habituation. In: Horn G, Hinde RA, editors. Short-Term Changes in Neural Activity and Behavior. Cambridge: Cambridge University Press, 1970:3–40. [17] Hull CL. Learning: II. the factor of the conditioned reflex. In: Murchison C, editor. A Handbook of General Experimental Psychology. Worcester, MA: Clark University Press, 1934:382 – 455. [18] Humphrey G. The Nature of Learning in its Relation to the Living System. New York: Harcourt Brace, 1933. [19] Kandel ER. Cellular Basis of Behavior: An Introduction to Behavioral Neurobiology. San Francisco: W.H. Freeman, 1976. [20] Kandel ER. Behavioral Biology of Aplysia. San Francisco: W.H. Freeman, 1979. [21] Kandel ER, Spencer WA. Cellular neurophysiological approaches in the study of learning. Physiol. Rev. 1968;48:65 – 134. [22] Kimble GA. Hilgard and Marquis Conditioning and Learning. Rev. edn. New York: Appleton-Century-Crofts, 1961.

.

[23] Lorenz KZ. The Foundations of Ethology. New York: SpringerVerlag, 1981. [24] Lubow, R.E. Latent inhibition. Psychol. Bull. 1973;79:398–407. [25] Mackey SL, Glanzman DL, Small SA, Dyke AM, Kandel ER, Hawkins RD. Aversive stimuli produce inhibition as well as sensitization of the siphon withdrawal reflex of Aplysia: a possible behavioral role for presynaptic inhibition mediated by the peptide FMRFamide. PNAS 1987;84:8730 – 4. [26] Mackintosh NJ. Conditioning and Associative Learning. Oxford: Oxford University Press, 1983. [27] Marlin NA, Miller RR. Associations to contextual stimuli as a determinant of long-term habituation. J. Exp. Psychol. (Anim. Behav.) 1981;7:313 – 33. [28] Pavlov IP. Conditioned Reflexes. London: Oxford University Press, 1927. [29] Peeke HVS, Petrinovich L. Approaches, constructs, and terminology for the study of response change in the intact organism. In: Peeke JVS, Petrinovich L, editors. Habituation, Sensitization, and Behavior. Orlando: Academic Press, 1984:1 – 16. [30] Petrinovich L, Widamen KF. An evaluation of statistical strategies to analyze repeated-measures data. In: Peeke JVS, Petrinovich L, editors. Habituation, Sensitization, and Behavior. Orlando: Academic Press, 1984:155 – 204. [31] Pilz PKD, Schnitzler H. Habituation and sensitization of the acoustic startle response in rats: amplitude, threshold, and latency measures. Neurobiol. Learn. Mem. 1996;66:67 – 79. [32] Rakitin A, Daniel T, Maldonado H. Habituation and sensitization to an electrical shock in the crab Chasmagnathus. Effect of background illumination. Physiol. Behav. 1991;50:477 –87. [33] Randich A, LoLordo VM. Associative and non-associative theories of the UCS preexposure phenomenon: implications for Pavlovian conditioning. Psychol. Bull. 1979;86:523 – 48. [34] Rescorla RA. Pavlovian conditioning and its proper control procedures. Psychol. Rev. 1967;74:71 – 80. [35] Rescorla RA. Behavioral studies of Pavlovian conditioning. Annu. Rev. Neurosci. 1988;11:329 – 52. [36] Romano A, Lozada M, Maldonado H. Nonhabituation processes affect stimulus specificity of response habituation in the crab Chasmagnathus granulatus. Behav. Neurosci. 1991;105:542– 52. [37] Sheafor P. ‘Pseudoconditioned’ jaw movements of the rabbit reflect associations conditioned to contextual background clues. J. Exp. Psychol. (Anim. Behav.) 1975;104:245 – 60. [38] Skinner BF. The Behavior of Organisms: An Experimental Analysis. New York: Appleton, 1938. [39] Teyke T. Food-attraction conditioning in the snail Helix pomatia. J. Comp. Physiol. A 1995;177:409 – 14. [40] Thompson RF, Groves PM, Teyler TJ, Roemer RH. A dualprocess theory of habituation: theory and behavior. In: Peeke HVS, Herz MJ, editors. Habituation. vol. 1. Behavior Studies. New York: Academic Press, 1973:239 – 272. [41] Thompson RF, Spencer WA. Habituation: a model phenomenon for the study of neuronal substrates of behavior. Psychol. Rev. 1966;73:16 – 43. [42] Thorpe WH. Learning and Instinct in Animals. London: Methuen, 1963. [43] Thorndike EL. The Psychology of Learning. New York: Columbia University Press, 1913. [44] Whitlow JW, Wagner AR. Memory and habituation. In: Peeke HVS, Petrinovich L, editors. Habituation, Sensitization, and Behavior. Orlando: Academic Press, 1984:103 – 153. [45] Wickens DD, Wickens CD. Some factors related to pseudo-conditioning. J. Exp. Psychol. 1942;31:518 – 26.