Mallea, J., Bustamante, J., Miguez, G., & Laborda, M. A. (2019). Classical Conditioning.

C
J. Vonk & T. Shackelford (Eds.), Encyclopedia of animal cognition and behavior. Cham,
Switzerland: Springer International Publishing.

Classical Conditioning much more than only simple reflexes and that it
is a complex process involving complex
Jorge Mallea1, Javier Bustamante2, responses. However, the basic concepts presented
Gonzalo Miguez3 and Mario A. Laborda3 by Pavlov have remained in the vocabulary of
1
Department of Psychology, Columbia researchers and the psychologists interested on
University, New York, NY, USA associative learning.
2
Institute of Social Sciences, Universidad de A CS is a stimulus that, after training, predicts
O’Higgins, Rancagua, Chile that the US will occur, usually causing an antici-
3
Universidad de Chile, Santiago, Chile patory CR. The US, by contrast, is a stimulus that
does not need training to elicit a response and is
usually relevant to the organism (Pavlov 1927).
Introduction CSs and USs are often called by other names
which allude to the same concepts. The CS is
Classical or Pavlovian conditioning is a type of also referred sometimes as a cue or a predictor.
learning where two or more events of the environ- The US is often referred as the outcome or the
ment are associated. This type of learning helps consequence. All these names are used somewhat
organisms to organize their behavior and repre- indistinctively from another.
sent their world. In a classic experiment, Pavlov Classical conditioning is crucial for animals to
(1927) discovered that a dog would salivate to the be prepared for or anticipate events of the envi-
presence of a sound if this sound was previously ronment, and to predict what will happen around
presented contiguously with food powder. Pavlov them. Consequently, it plays an important role for
called the food powder an unconditioned stimuli animals to adapt to their environment (Domjan
(US) that evoked a response by its own, which he 2005). The next sections will review how the
called an unconditioned response (UR). The view of classical conditioned has changed from
sound was originally a neutral stimulus that did its discovery, the basic preparation and phenom-
not produce any relevant response, but after being ena of classical conditioning studied in the labo-
presented with the food, the sound became a con- ratory, and some of the most influential theoretical
ditioned stimulus (CS), which elicited a condi- developments in the area. We will also cover how
tioned response (CR). To Pavlov, this response classical conditioning helps individuals to adapt
was “conditional” to the unconditioned stimulus, to their environments, some of its main applica-
thereby the name. The study of classical condi- tions in therapeutic contexts, and its current state
tioning has changed since Pavlov’s years; we now as theory and tool in Psychology.
know that classical conditioning is involved in
© Springer Nature Switzerland AG 2019
J. Vonk, T. K. Shackelford (eds.), Encyclopedia of Animal Cognition and Behavior,
[Link]
2 Classical Conditioning

Historical Background conditioning (for a detailed depiction of experi-

mental preparations, see the corresponding sec-
Pavlovian or classical conditioning was discov- tion), also extended to different species including
ered independently in Russia and in the USA, at rabbits, rats, and humans. Early learning models
the end of the nineteenth century. Edward also were developed during this time, for instance,
Twitmyer, at the University of Pennsylvania, Clark Hull’s Principles of Behavior (1943) and
investigated whether the patellar tendon or Edward Tolman’s influential Purposive Behavior
“knee-jerk” reflex worked under facilitating con- in Animals and Men (1951). Theoretical models
ditions. The sound of a bell signaled with 500 ms during this period followed closely the original
of anticipation the knee tapping. Shortly after- Pavlovian view of classical conditioning. Primar-
wards, Twitmyer observed that participants ily, the predominant conception was that condi-
extended their knees immediately after the sound tioning was an automatic, low-level process
of the bell but in absence of any tap to the knee. which involved mostly smooth muscle tissue and
Once it was demonstrated that participants were glandular responses such as those studied by Pav-
not doing it intentionally, Twitmyer concluded lov, and that the key factor that determined how
that some stimulus other than the usual one had two stimuli are associated was temporal contigu-
elicited the reflex. ity. This view shifted dramatically during the sec-
Twitmyer presented his findings at the Annual ond half of the twentieth century as the result of
Meeting of the American Psychological Associa- new evidence (e.g., blocking effect; Kamin 1969),
tion with little impact, and he did not extend his which suggested to learning theorists that classical
work on conditioning. Thus, its discovery is conditioning was a more complex and rich learn-
credited to Ivan Pavlov, the Russian physiologist ing process. For example, Rescorla (1988) sum-
who worked approximately at the same time on marized classical conditioning as “the learning
conditioning in his laboratory, and upon realizing that results from exposure to relations among
what he had found, dedicated the following events in the environment. . .a primary means by
decades of his life to the study of classical which the organism represents the structure of its
conditioning. world” (p. 152).
Pavlov’s work with conditioned responses In parallel with this theoretical and empirical
became widely known during the first decades of developments, there was also an important devel-
the twentieth century, particularly in the USA. For opment of clinical applications based on associa-
instance, as early as 1914 John B. Watson and his tive principles. For example, the work of John
student Karl Lashley were conducting studies in B. Watson, Rosalie Rayner, and Mary Cover
classical conditioning at Watson’s laboratory. Jones showed how classical conditioning was
Watson advocated in favor of the conditioned able to explain the acquisition of fear as well as
reflex as an explicative mechanism for many psy- evaluate techniques to treat it. Later the South-
chological processes and was heavily influential African psychiatrist Joseph Wolpe developed the
in its adoption by American psychologists. The systematic desensitization procedure to treat anx-
publication of Conditioned Reflexes in 1927 iety, applying what was learned in the experimen-
(translated by G. V. Anrep, former student of tal laboratory to real life situations. These
Pavlov) further helped increase the interest in instances laid the ground for an empirical
classical conditioning. approach to psychotherapy in general and partic-
During the first half of the twentieth century, ularly for behavior therapy. The translation from
classical conditioning was widely studied as one basic experimental research to clinical application
of the fundamental learning processes (the other still continues today, with a growing corpus of
one being instrumental or operant conditioning). empirical results in many areas.
Basic experimental paradigms were extended As a result of the empirical and theoretical shift
from the original Pavlovian design and included mentioned above, research in classical condition-
preparations still in use today, such as fear ing has growth beyond its origins and has
Classical Conditioning 3

included novel experimental paradigms which experimental preparation), subjects will respond
came to be seen as associative in nature (e.g., to the CS in a similar way to how they respond to
contingency judgment and causal learning; Dick- the US, or, depending on the task, with a response
inson and Shanks 1985). An important develop- related to the specific features of the CS.
ment also occurred in models of associative Several parameters affect the acquisition of a
learning (e.g., Mackintosh 1975; Rescorla and CR and thus will vary across experimental prepa-
Wagner 1972). The mathematical formalization rations. First, more intense or salient stimuli are
of learning models provided testable predictions generally associated with faster and stronger
which further helped bolster research in classical acquisition (Pavlov 1927). Acquisition also pro-
conditioning, in a trend that continues today. This ceeds faster when the stimuli are novel, as shown
expansion in the study of classical conditioning by the “pre-exposure” or “latent inhibition” effect
was also notorious in what is called the “func- (Lubow and Moore 1959). Finally, acquisition is
tional analysis” in contemporary research (e.g., also affected by the temporal relationship between
Domjan 2005; Hollis 1997), an approach that the CS and the US (inter-stimuli interval) and
focused on investigating the role and properties between each trial (inter-trial interval; e.g.,
of classical conditioning as an evolutionary adap- Gallistel and Gibbon 2000).
tive process. Contiguity and contingency are two key con-
Currently, classical conditioning is an active cepts in Classical conditioning. Contiguity refers
research field, with relevant and growing links to the spatial and/or temporal relationship
with the contemporary neuroscience and cogni- between the CS and the US. Contingency is the
tive psychology, and particularly relevant as a relationship of probability between the CS and the
model of information processing (for more US, that is, how likely it is that both stimuli are
details, see the final section of this entry). presented together compared to the likelihood of
them presented alone.
Variations in contiguity have an impact on the
Major Concepts effectiveness of conditioning, and thus have been
extensively investigated. The most used proce-
This section will review some relevant basic dure is the so-called delay conditioning, in
aspects of Classical conditioning, such as Pavlov- which the US begins after the onset of the CS
ian acquisition, excitation, extinction and inhibi- but overlaps with it. This delay between the onsets
tion, and relevant factors such as contiguity and of both stimuli can be short or long. In “trace”
contingency. conditioning, there is a time gap between the
The process by which the CR comes under offset of the CS and the onset of the US. The
control of the CS is called acquisition. This con- association in this case is supposed to occur
trol is usually acquired by the CS after being between a trace of the already absent CS and the
paired with the US. The CS is generally a stimulus US, hence the name. Both stimuli can also be
that by itself and prior to any training does not presented simultaneously. Finally, in backward
elicit the target response. Such target response is conditioning, the usual temporal relationship is
typically similar to that elicited by the inverted, and the US is presented before the CS.
US. Because of this, the CS is usually viewed as Contiguity has had theoretical value as an
more neutral than the US, and this relative differ- explanation of Classical conditioning. For
ence is used to define what the CS and the US are instance, for Pavlov (1927), contiguity was the
in any given experiment. For example, a weak most important element for an association to
electric shock can be used as US in an experiment, occur. Past and current views about the role of
and as CS in a second one, provided that the contiguity in conditioning are discussed in detail
corresponding CS and US are weaker and more in the corresponding section.
intense, respectively. After one or more pairings Contingency, on the other hand, is classified in
(the specific number will vary depending on the three variations depending on the relative
4 Classical Conditioning

probability of the CS and the US. Contingency procedure (Rescorla 1969), the CS is negatively
can take positive, negative, and zero values. In correlated with the US. In this case, the excitatory
positive contingency, the US is more likely to expectancy is supposed to be given by the exper-
occur when the CS is presented than when it is imental context. Because the US is presented con-
not; in negative contingency, the US is less likely sistently in the presence of signals provided by the
to occur when the CS is presented than when it is experimental context, it should become excitatory
not; and in zero contingency, the probability of the while the CS becomes inhibitory.
US is the equivalent in the presence of the CS and Conditioned inhibition is supposed to be con-
in its absence. trary to excitation in that while the latter increases
The idea of contingency has been relevant for CRs, inhibition does not, which makes measuring
modern conceptualizations of Classical condition- it more difficult. Several techniques have been
ing. Following the work of Rescorla (1969), con- developed to overcome this problem:
tingency is thought to reflect the informational
value of the CS, that is, whether the CS gives (a) Bidirectional responses: Some responses can
any meaningful information about the occurrence increase or decrease compared to a baseline.
of the US. Thus, in positive contingency, the CS is Typical examples are the approach/avoidance
a reliable predictor of the US, while in negative system, or physiological responses such as the
contingency, the CS is a predictor of the absence heart rate. A bidirectional response may, for
of the US, and in zero contingency, the CS does example, decrease in presence of an inhibitor
not give any information about the occurrence of and increase in the presence of an exciter.
the US. Different models give alternative interpre- Nevertheless, some responses have a very
tations for each case, and thus the role of contin- low baseline, making this type of measure-
gency for models of Classical conditioning is ment not ideal.
further developed in the corresponding section. (b) Summation/Retardation tests (Rescorla
The variations in the informational value of the 1969): The summation test consists of the
CS regarding the US are closely related to the presentation in compound of an excitatory
concepts of excitation and inhibition. Excitatory CS+ and an inhibitory CS-. If the CS- is
conditioning refers to when the CS is a good effectively a conditioned inhibitor, it should
predictor of the US as in positive contingency. decrease the response elicited by the CS+,
Conditioned inhibition, on the other hand, refers compared to presentations of the CS+ alone.
to when the CS is a predictor of the absence of the In the retardation test, the CS- is paired with
US, as in negative contingency. the US. If the CS- did acquire conditioned
A key feature of conditioned inhibition is that it inhibition, this reacquisition training should
occurs when there is a previous expectancy of the proceed slower compared to a novel CS.
US. In other words, the organism must somehow
expect the occurrence of the US so that the inhib- After acquisition, repeated presentations of the
itory CS can predict its absence, that is, there must CS alone cause a gradual decrement in the condi-
be at least a previous excitatory context. The pro- tioned responding, which can eventually lead to
cedures designed to train conditioned inhibition its complete suppression. Both the procedure and
all include a previous expectancy of the occur- its result are called extinction. Pavlovian extinc-
rence of the US. In the Pavlovian or standard tion has been widely researched in the last decades
conditioned inhibition procedure (Pavlov 1927), for the study of its mechanisms and clinical appli-
in an initial phase subjects receive CS1-US trials cations (mainly for exposure therapy, details, see
for the CS1 to provide an excitatory context. In a the “Applications Outside of the Laboratory”
second phase, CS1 is presented together with a section).
CS2 without the US. With this procedure, CS2 One relevant feature of extinction is that its
should become inhibitory, since it signals the effect on behavior is usually not permanent, and
absence of the US. In the negative contingency the response to the CS recovers under several
Classical Conditioning 5

circumstances. Renewal is the recovery after the used to measure the amount of learning about the
subject is moved to a different context to the one now aversive CS.
in which extinction was conducted (e.g., Bouton In most of the studies, both with human partic-
and Bolles 1979). There are three types of ipants and non-human subjects, fear conditioning
renewal, depending on the procedure. ABA is examined with the pairing of a CS or context
Renewal occurs when acquisition is conducted and a mild electric shock. The shock is regulated
in a Context A, extinction in a Context B, and in its intensity by the participants themselves in
the subject is then tested in the acquisition con- the case of humans, to be aversive without being
text. ABC Renewal is similar, although test is physically harming. Fear is indirectly measured
conducted in a different Context C. Finally, in by several physiological and behavioral
AAB Renewal both acquisition and extinction responses. In the case of human participants,
are conducted in Context A while testing is these measures include skin conductance or gal-
conducted in a different context. Spontaneous vanic skin response, pupillary dilation, and, par-
recovery is the response recovery that happens ticularly in the case of exposure therapy,
when a time interval is introduced between extinc- approach/avoidance responses. In rats, freezing
tion and testing (Pavlov 1927). In reinstatement is usually reported as a measure of fear. Freezing
(Rescorla and Heth 1975), the response recovers can be measured as the time the rat stays without
when the US is presented again after extinction. moving, which can be recorded and then exam-
Finally, when subjects receive again CS-US ined by independent judges or by how much it
pairings after extinction, response recovers faster disrupts the rat’s normal behavior. Similar is the
compared to the acquisition of a newly trained procedure of conditioned emotional response or
association (rapid reacquisition; Pavlov 1927). “conditioned suppression.” In this procedure, the
rat is initially trained to press a lever for food. In a
second phase, the subject receives CS-shock
pairings. During testing the rat has access again
Experimental Preparations
to the lever, and lever press is measured in pres-
ence of the CS and in its absence. Fear is indicated
This section will focus on experimental prepara-
by the decrement in lever pressing in the presence
tions used in Pavlovian research such as fear con-
of the fear-eliciting CS compared to the baseline
ditioning, taste aversion, and others. One
responding without the CS. There are several var-
important issue about Classical conditioning is
iations in which the effect of fear is measured
the question regarding the generality of Pavlovian
against different behaviors such as licking water
phenomena. Pavlov himself (e.g., Pavlov 1927)
tubes or magazine entries for pellets, although the
studied conditioning using stimuli of different
basic principle is the same.
modalities (i.e., visual cues, tones, tactile stimuli)
A suppression ratio can be calculated as an
to examine whether conditioning occurred or was
index for the suppression of behavior. The sup-
similarly effective in all cases. The following
pression ratio considers how much responses
years saw the development of many different
occurred during the CS relative to the sum of
experimental preparations, using CSs and USs of
those responses and the baseline responses regis-
different modalities, and different ways to mea-
tered before the onset of the CS. The suppression
sure the conditioned response.
ratio ranges between 0 and 0.5, where 0 indicates
Fear conditioning: Fear conditioning is one of
no response and thus complete disruption of
the most studied experimental preparations in
behavior (i.e., fear) and 0.5 indicates no fear.
Classical conditioning. It consists of the pairing
Conditioned taste aversion: Classical condi-
of a given CS with an aversive or fear-eliciting
tioning plays a relevant role in the acquisition of
US. After training, the CS comes to elicit
taste preferences and aversions (Garcia and
responses (i.e., anxiety and/or avoidance) origi-
Koelling 1966). In a conditioned taste aversion
nally elicited by the aversive US, which can be
experiment, subjects (typically rats) receive a
6 Classical Conditioning

flavor-CS (i.e., food or liquid) followed by sick- the US: goal-tracking would be more likely when
ness provoked by administering lithium chloride the CS and the US are separated temporally and
or other emetic procedure. Conditioned taste aver- spatially, while sign-tracking would be more
sion can be acquired after a single CS-US pairing, probable when both stimuli are close.
which is typical in aversive conditioning, but not Sign-tracking, and particularly autoshaping,
observed in appetitive paradigms. Also, subjects has been historically a relevant phenomenon for
may present taste aversion even if hours have classical conditioning. From the early conceptu-
passed between the presentation of the flavor-CS alizations of Pavlov, it was considered that classi-
and the internal sickness. These features have cal conditioning affected reflexes and glandular
been of theoretical relevance for discussing the responses, while responses involving skeletal
role of preparedness or pertinence of the stimuli muscles were the domain of operant conditioning.
in Pavlovian learning. Sign-tracking, as primary classical conditioning
Eye-blink conditioning: The eye blink is a procedure that involves complex skeletal
reflex common to a wide variety of species. Any responses, showed that this distinction was not
object approaching the eye, such an airpuff, will empirically valid (Hollis 1997).
elicit the reflex response. Eye-blink conditioning Drug tolerance: Drug tolerance is the decre-
thus works by pairing a visual or auditory cue with ment in the effect of a drug after its repeated
a reflex-eliciting stimulus, usually an air puff or a administration or, conversely, the need of higher
mild electric shock (Gormezano 1966). Eye-blink doses to obtain the same original effect. Drug
conditioning has been extensively researched tolerance has a learning component in what is
because of its simplicity in humans and other called “associative tolerance.” Research has
animals. It has also been relevant for understand- shown that each episode of drug administration
ing basic neural mechanisms of learning and is a conditioning trial, in which the effect of the
memory. drug is associated with, and thus predicted by, a
Sign and goal tracking: Sign- and goal- wide arrange of stimuli (Siegel et al. 2000). For
tracking are terms that refer to similar tasks and instance, the presence of friends or bottles can act
include related preparations as autoshaping as a signal for the consumption of alcohol. Within
(Brown and Jenkins 1968). In sign-tracking, a this framework, tolerance is the product of antic-
visual CS is paired with an appetitive US, such ipatory conditioned responses to these signals that
that after repeated pairings, the animal responds predict the drug and thus allow the organism to
by approaching and contacting the CS. In a typical adapt to its effect. For example, a well-known
experiment, a pigeon is trained to associate a effect of alcohol is the decrement in body temper-
keylight (i.e., an illuminated disc) with the deliv- ature. The compensatory reaction by the organism
ery of grain. After some trials, the pigeon will is to rise the temperature to counteract the effect of
approach the keylight instead of the food maga- alcohol intake. This response is specific to the
zine and peck it as if it was a pellet of grain. In cues or contexts associated with the drug, that is,
goal-tracking, the basic procedure is similar associative tolerance will be evidenced only in the
although the measured response is the approach specific circumstances linked to usual drug con-
to the place of appearance of the US. sumption where the drug-predictive cues are
Since both tasks are similar, some research has present.
examined the factors that may explain whether Causal learning: Causal learning is a broad
organisms will respond with either sign- or goal- term that encompasses different methods by
tracking. For instance, it has been shown that rats which people learn about the causal structure of
tend to show more goal- than sign-tracking, that a given task. One of them is the so-called human
is, rats approach the food magazine instead of the contingency learning, although the terms “predic-
CS, while pigeons show the opposite pattern. It tive learning” and “causal judgment” are also used
has also been suggested that the observed behav- in the literature. Participants, in a contingency
ior depends on the contiguity between the CS and learning task, are presented with sets of stimuli
Classical Conditioning 7

and then asked to assess the causal relationship two CSs are presented simultaneously followed
between them. For example, in what is perhaps the by the US. The critical feature is that one of the
most used causal learning task, participants assess CSs is defined as more salient than the other (i.e.,
the causal relationship between meals or more intense or relevant). When both stimuli are
chemicals and sickness in a hypothetical patient. tested, subjects show more conditioned response
In such tasks there are many ways of assessing in the presence of the more salient CS compared to
this relationship, from simply giving a prediction the less salient one.
in an arbitrary scale to more behavioral of physi- Potentiation (Rusiniak et al. 1979) is a varia-
ological measures like response latency and eye tion of the overshadowing procedure frequently
tracking. Contingency learning shows many of observed in taste aversion experiments. The pro-
the phenomena found in classical conditioning, cedure is similar to overshadowing training, but
which is why it is usually explained in terms of the result is the opposite, that is, the less salient CS
associative learning models (Dickinson and elicits more conditioned response than the more
Shanks 1985). For instance, acquisition and salient CS during testing. Some factors have been
extinction of a causal relationship proceed in a suggested to facilitate potentiation over over-
similar manner to other Pavlovian preparations, shadowing: the trace between CS and US, the
as do cue competition effects. specific experimental preparation (most of the
Conditioned Analgesia: Stressful situations observations of potentiation have been conducted
can produce an unconditioned reduction in the in taste aversion), and the relative salience of the
severity of pain. Consequently, an initially neutral trained CSs.
stimulus, through pairings with a stressor, also Blocking is perhaps the most investigated cue
comes to produce a reduction in the response to competition effect. In blocking (Kamin 1969),
painful stimulation. Such reduction is often called subjects receive an initial phase of CS1-US until
conditioned analgesia (i.e., conditioned stress- the conditioned response to CS1 is reliably
induced analgesia). This preparation has been established. In a subsequent phase, subjects are
used both with human and non-human animals, trained with CS1-CS2-US pairings, where CS2 is
who typically report or show behavior indicative a second, different CS. When both CSs are tested
of lower pain when the associated CSs are pre- individually, subjects respond more to CS1 than to
sented (see Miguez et al. 2014, for a review). CS2, that is, CS2 is “blocked” by the CS1.
Interference: In Classical conditioning, it is
frequently observed that two different indepen-
Key Phenomena dent trainings that share a common element can
affect how subjects respond to the stimuli
Several learning phenomena have been histori- involved in each training. This general phenome-
cally relevant for the study of Classical condition- non is called “interference” (Bouton 1993).
ing and for the development of theoretical models. Pavlovian interference can be examined in two
This section describes some of them, such as cue different ways. In proactive interference, a first-
competition and interference phenomena. learned association can affect responding to a
Cue Competition: Cue or stimulus competition second-learned association; in retroactive interfer-
refers to a set of Pavlovian phenomena, in which ence, a first-learned association can be affected by
multiple CSs are trained simultaneously with an a second-learned association.
outcome or US. In these cases, all CSs involved Furthermore, interference also varies
appear to “compete” with each other for the con- depending on the common element shared by the
ditioned response, thus the name. When both different associations; cue-interference refers to
stimuli are tested individually, one will elicit the training in which the first- and the second-
less-conditioned response than the other one. learned association share a common US but dif-
Overshadowing is a cue competition effect first ferent CSs. For example, subjects trained with
identified by Pavlov (1927). In overshadowing, tone CS-shock pairings in a first phase, and light
8 Classical Conditioning

CS-sock pairings in a second phase will show less change on learning) would be progressively
fear to the light compared to subjects that received smaller as learning progresses.
only one of either phase of training. These ideas were taken by Robert Rescorla and
Outcome-interference refers to different asso- Allan Wagner to create one of the most influential
ciations sharing the same CS but different USs or models of learning (Rescorla and Wagner 1972).
the absence of a US as in extinction. For example, This model states that learning is the difference
in the case of counterconditioning, a tone CS can between the maximum possible amount of learn-
be initially paired with a shock, and in a second ing about an outcome and the amount of informa-
phase with food pellet. Another example is latent tion supplied by all the cues present in the past
inhibition, in which subjects are presented to a CS pairings with the US.
alone (without a US or consequence) during a first Under this conception, learning could also be
phase, and then to CS-US pairings during a sec- considered a function of the difference between
ond phase. Such subjects acquired excitatory con- what happens and what the subject predicts that
ditioning slower relative to subjects that did not will happen. Rescorla and Wagner also proposed
experienced the first phase. that learning depended on the product of how
salient or intense the cues and the consequences
are, which, according to them, would not change
Theoretical Developments during the course of learning. The result of the
equation is a quantitative measure of learning that
Researchers and theorists of learning have created they called associative value (V).
several accounts to explain how stimuli are asso- The Rescorla and Wagner (1972) model was
ciated and what happens when we observe the first to compute the prediction from multiples
blocking, inhibition, or any other conditioning CSs simultaneously; consequently, it explains
phenomena. These accounts are usually referred several of the competition phenomena observed
as theories or models of conditioning or learning. in Classical conditioning, like blocking. For the
The first theory of conditioning is as old as model, in the first phase of training, when the first
conditioning itself. Pavlov (1927) thought that cue is paired with US, the cue becomes a good
during the presentation of the US, a node in the predictor of the US, but when the second stimulus
brain was activated, which was linked to a is added in the second phase, this second cue does
response node for the unconditioned response. not provide much more information about the US:
When the CS started to be paired with the US, the subject is already good in predicting what is
the CS-node began to supplant the US-node, going to happen. In terms of the model, at the end
thereby producing the same response that the US of the pairings of the first cue with the US there is
produced before. This early theory is not able to little to no difference between what happens and
explain all the phenomena and contemporary evi- what the subject predict that will happen, and
dence on conditioning but serves as a first expla- when the second CS is introduced, there is little
nation of how animals learn new conditioned or no more to learn.
reflexes linked to innate responses. The principles of the Rescorla-Wagner model
Most of the currents models of learning are have been taken by several other learning theo-
influenced by proposals from Clark C. Hull, his rists, helping to create many of the current models
disciple Kenneth Spence, and Bush and Mosteller. of conditioning and generating research until
In different ways, they all suggested that learning today to test the assumptions and predictions of
is a function of a maximum amount of possible the model.
learning, and what or how much the subject has Mackintosh (1975) and Pearce and Hall (1980)
learned at that moment. This rule assumes that took the basic principle of the Rescorla-Wagner
learning is a gradual process based on error cor- model but added, in two different ways, a new
rection, that is, that the amount of learning gained assumption. They assumed that the salience or
each time that the stimuli are paired (the rate of intensity of the CS, what they called
Classical Conditioning 9

“associability,” would not remain constant across consequence nothing. Here the absence of a stim-
learning trials and it would vary proportionally to ulus can also serve as consequence after the
how good of a predictor the CS becomes. These CS. This also causes the attention to the cue to
models are commonly referred to as “attentional” decrease but by a different mechanism than for
models, because of the shared characteristic that Mackintosh’s model. The result in both cases is
the subject will attribute more importance or the same, the subject will take more trials to learn
salience to cues based on their history with the the CS-US association due to the low associability
consequences. caused by pre-exposure, but produced by different
Mackintosh (1975) proposed that cues that are mechanisms.
better predictors will have a higher associability or Recent developments have tried to reconcile
attention. If a stimulus predicts a consequence, it this two, seemingly opposed, proposals creating
should be considered more relevant and thereby models that assume that the associability of stim-
should be paid more attention. Due to this, sub- uli can increase in both cases, by being a good
jects will learn faster associations involving cues predictor of the US and by being an inconsistent
that are better predictors of consequences. predictor of the consequence. These theories are
Pearce and Hall (1980) model has a seemingly known as “hybrid” models and have been highly
opposed approach to Mackintosh’s to the problem discussed in the last years (e.g., Pearce and
of associability. They proposed that subjects do Mackintosh 2010).
not need to pay attention to cues that are already A somewhat different approach was taken by
known, and are good predictors. Instead, subjects Wagner (1981) who was interested in how repre-
should pay attention to new cues or cues that are sentations are formed in memory with his theory
inconsistent predictors of consequences. For of standard operating procedures (SOP). For this
Pearce and Hall, stimuli that are not good pre- theory, stimuli consist of nodes of a number of
dictors of the CS, for example, before the begin- elements. These elements and by in three different
ning of training, will have a higher associability or states: an inactive state, and two active states
attention, and as the pairings with the conse- (A1 and A2). Stimuli are normally inactive, but
quence continue, the associability will become would activate in A1 when they are directly pre-
smaller and smaller. According to the model, sented. Stimuli can then be activated in A2 either
learning about new cues when they are paired by normal decay from A1 after some time or after
with the US should be faster than cues that are learning when an associated cue is presented.
already good predictors of the US. After a stimulus is in A2, it can only move to the
The evidence shows that both Mackintosh inactive state. Excitatory conditioning occurs
(1975) and Pearce and Hall (1980) models cor- when the two stimuli are in A1, and as learning
rectly predict how subjects behave in various cir- progresses, the first stimulus starts to activate
cumstances. Both models are able to explain more and more elements of the second in A2.
conditioning phenomena such as blocking or This theory has been fundamental to explain phe-
latent inhibition. For example, the Mackintosh nomena like context specificity of latent inhibi-
model states that in latent inhibition, when the tion, and to make predictions about learning in a
CS is presented in absence of any consequence, real-time base instead of trials as the Rescorla &
the subject pays less attention to it because the cue Wagner model.
is not a good predictor of the consequence; in the Other theories of learning have put focus on
following phase, during the presentations of the how the US is processed by the subject during
cue with the outcome, subjects will take more time learning. These theories assume that, in any inter-
to learn the relation between them compared to a action with the environment, the target US is not a
cue that was not pre-exposed. Pearce and Hall’s simple stimulus defined by just one dimension or
model, on the other hand, assumes that when the attribute. Consequences are always complex and
CS is pre-exposed, the cue becomes a good pre- involve many possible characteristics from which
dictor of a specific consequence, being that to learn. For example, if the presence of a dog is
10 Classical Conditioning

followed by its bite (possibly causing fear to it), theories of learning have considered how
this bite is not just composed by the pain itself, but response changes independently of acquisition
it can be divided in different parts like the vision training, and how learning can occur without an
of the bite, the motivational state that it causes apparent change in response. Miller and Matzel
(probably non-rewarding), the time of the bite (1988) provided an approximation to this in the
(when the bite occurs both in a general timeline “comparator hypothesis.”
and since the dog was perceived), the response Miller and Matzel (1988) suggested that when
that it causes, and the emotional reaction, among the subject learns, it does not take in consideration
others. Some theories have proposed that we can how good of a predictor a cue is (as the Rescorla-
learn from all these attributes separately and that Wagner model); instead, the organism learns
the rules to learn the relationship between the cue about all the stimuli that are presented together.
and the consequence may be different for each of In other words, contiguity alone is sufficient to
its parts (e.g., Delamater and Oakeshott 2007). produce learning. Since the subject learns about
Another family of theories views animals as all present stimuli, responding must be controlled
able to learn about the complex environment by a different process than the contingency or the
when several stimuli are in place, and not only associability of each cue. Because the subject is
about its elements. Each stimulus can be seen as a learning via contiguity about every present stim-
part of a more complex structure in the subject’s ulus, the model suggests that the organism com-
environment. Pearce (1987) proposed a model pares which one of all the present stimuli has the
that elaborates on these assumptions. The model strongest connection with the US. When a cue is
states that subjects do not learn about each stimu- present, subjects will compare the CS with other
lus separately, but at the moment of learning, they cues that were present during training; if the cue is
process the environment as a unique complex the more strongly associated with the conse-
stimulus composed of every single present cue. quence, the subject will respond. Conversely, if
Thus, individuals do not learn about relationships other cues have a stronger relation with the US
between a single cue and the US but about this than the CS being presented, subjects will not
complex cue (called “configurational stimulus” or respond to that CS.
simply “configuration”) and how that stimulus is The comparator hypothesis (Miller and Matzel
related to the consequence. 1988), more specifically, states that in every learn-
The Pearce (1987) model states that, during ing situation (when a cue is followed by a conse-
testing, the more similar the configurational stim- quence or US), the cue is associated with the
ulus presented is to the one encountered during consequence, but this cue is also associated with
training, the more the subject will respond to it; any other stimulus presented during this learning
conversely, the less similar the test configuration procedure, which causes that when the cue is
to the original, the less response will be observed presented, the subject activates a memory not
(effect known as “generalization decrement”). only of the consequence but also of all other
This model is able to explain several learning stimuli that were present during training. This
phenomena, including cue competition and memory in turn also indirectly activates a memory
others. of the consequence. When a cue is presented, the
The theories presented above all assume that subject compares which memory of the conse-
subjects respond as much as they learn. In other quence is stronger, the directly activated by the
words, they assume that the response by the sub- cue presented or the one activated indirectly via
jects is proportional to the acquired learning stated the memory of the other stimulus presented during
in the equation, which changed during training or testing. Due to this mechanism, the model predicts
acquisition. However, under many circumstances, that once individuals learn, they can change
organisms change their response without further responses without further training being neces-
training. The opposite also occurs and is possible sary; for instance, if a good competitor that was
to learn without any change in responding; blocking the response to another stimulus stops
Classical Conditioning 11

being a good competitor, the subject may start chances to survive. This suggests that classical
now responding to the previously blocked stimu- conditioning is not only a cognitive tool but that
lus. In this situation, acquisition relative to the also has a great evolutionary role in the species
blocked stimulus has not changed, just the rela- that learn through it.
tionship of the competitor stimulus with the CS. The different phenomena of classical condi-
Other family of models have set their focus on tioning and the mechanisms discussed in previous
the temporal relation between CSs and USs as a sections are general across all species were con-
central mechanism to learning. Models as the one ditioning is observed, and apply to diverse
proposed by Gallistel and Gibbon (2000) share response systems, such as fear, displeasure, pain,
with the comparator hypothesis the notion that taste, sex, etc. This suggests that conditioning has
responding to a stimulus depends on the compar- a generalized role in how animals represent their
ison between the target stimulus and the back- world and is not limited to certain kinds of learn-
ground stimuli. However, this model differs ing. One classical feature that shows the relevance
from the comparator hypothesis in that it assumes of classical conditioning in natural environments
that subjects encode temporal intervals between is that learning between cues that have a “natural”
events and rates of reinforcement instead of a relation to each other proceeds considerably faster
CS-US association. For this model, subjects will than learning between neutral or previously
respond to a stimulus when the rate of reinforce- unrelated stimuli. This natural relation between
ment of this stimulus is considerably higher than the stimuli can come from a CS that is part of a
the rate of reinforcement of the context. This and chain of actions that lead to the US, or of the CS
other time-focused models have been proved very being a feature of the US that initially does not
useful in many circumstances and have generated elicit a response; this last case is known as
a large cluster of research. “object” or “part-whole” learning (Domjan
All these models and theories here presented 2005). For example, learning to predict that a
have led to an immense increase in the under- certain food will cause sickness is considerably
standing of how organisms learn and what is easier if the predictor is the flavor of the food than
learned in classical conditioning. Moreover, the if it is the sound or vision of the food. The use of
different proposals have helped to conceive a naturally related stimulus in learning produces not
great amount of new research to test the models only a faster learning but is also more robust and
and account for their failures. Nevertheless, the resistant (Garcia and Koelling 1966). This evi-
current state of the art is likely far from having a dence implies an evolutionary predisposition to
single model that would be able to explain all the learn certain association faster than others, proba-
phenomena in classical conditioning. More than bly due to their importance and relative comple-
one hundred years of research have shown that mentarity between the stimuli. Even more,
classical conditioning is an extremely compli- animals learn to predict events that can be threat-
cated cognitive process, which mechanisms can ening to their lives (e.g., to predict predators, a
go from considerably simplistic to immensely painful stimulus, or poisoning) faster than less
complex in different situations and under different threatening events (e.g., to find a place, tolerance
perspectives. to a drug, or when food would come). From an
evolutionary perspective, not learning about the
former means a higher chance of dying than not
A Functional Point of View learning about the latter, which would create an
evolutionary predisposition to learn about danger-
Classical conditioning, as presented before, ous stimuli faster than to other kind of stimuli.
allows organisms to represent themselves their One example of conditioning with naturalistic
world. It allows subjects to act appropriately and stimuli with nonthreatening consequences is sex-
predict many situations in their natural environ- ual conditioning. In sexual conditioning, the sight
ment, improve their efficiency, and increase their of a female-like figure or an actual female is paired
12 Classical Conditioning

with access to copulation. Males that are subject to subjects with a task based on that of Shenger-
the pairings with the vision of an actual female as Krestinikova. In his study, a 6-year-old child had
CS are much faster in approaching the female and to learn discrimination between two different
ejaculating. As in previous examples, when the auditory stimuli, which were then progressively
naturalistic CS is used, learning is less susceptible made more similar to make the discrimination
to extinction and cue competition. more difficult. Both experiments are considered
Conventional descriptions of classical condi- as the first ones establishing that classical condi-
tioning emphasize the development of CRs to the tioning could be behind the acquisition of both
CS. However, numerous studies have shown that adaptive and nonadaptive emotional responses to
conditioning also alters how organisms react to, novel stimuli. These results were solidified and
and interact with, the US. This expands the defi- extended by the following research.
nition of classical conditioning to not only the Acquisition and elimination of fear: The next
way organisms represent their world but also a step was performed by John B. Watson, an Amer-
functional process through which animals ican psychologist and founder of behaviorism.
increase their chances of survival and procreation Watson conducted research aimed to demonstrate
(Hollis 1997). how principles of classical conditioning could
explain the acquisition and elimination of fear to
different objects. In what is perhaps the most
Applications Outside of the Laboratory known and infamous experiment in the history
of Psychology, he and his assistant Rosalie
The study of classical conditioning has been, and Rayner (Watson and Rayner 1920) conditioned
still is, relevant for the development and improve- an 11-month-old child called “little Albert” by
ment of applications outside of the laboratory. pairing a white rat, to which Albert showed pre-
This section lists and describes some of these viously no fear, with the sound produced by
applications in the area of Psychopathology and clanging iron rods behind Albert’s head. After
Psychotherapy. several parings, Watson and Rayner presented
Experimental neuroses in dogs and humans: the rat by itself, to which the boy showed fear.
Applications of classical conditioning began to Afterwards they examined whether fear could
be developed shortly after Pavlov’s initial demon- generalize to other objects by showing Albert a
strations of conditioning (Pavlov 1927). One of rabbit, a dog, and a fur coat, to which Albert also
Pavlov’s collaborators, Shenger-Krestinikova, is reacted with fear. Unfortunately, Watson and
cited as the first one who examined the so-called Rayner never had the opportunity to treat Albert’s
experimental neurosis in dogs. Shenger- newly acquired fear, but an attempt in this direc-
Krestinikova implemented a task in which a dog tion was made in a subsequent experiment with a
had to discriminate between two different visual child called Peter. This child showed fear to white
cues (i.e., a circle and an ellipse), one of which rats and rabbits, probably by means of condition-
was followed by food. Once the discrimination ing during his life. Mary Cover Jones (Jones
was established, indicated by the animal salivat- 1924), developed a procedure to reduce Peter’s
ing to the circle but not to the ellipse, both stimuli fear by exposing the child to the fear-eliciting
were gradually made more similar. The dog dis- objects in what they called “direct conditioning.”
criminated successfully between both visual cues Peter was put in a chair and given a snack while in
until they were practically indistinguishable, at parallel a white rabbit in a cage was put on the
which point the animal showed responses not table in a distance that did not disturb the child.
displayed previously which were, according to The procedure was repeated the next days with the
the researchers, indicative of emotional distress rabbit slightly closer each time, until Peter could
or akin to neurotic behavior. eat with the rabbit next to him, and even while
Nikolai Krasnogorski, another of Pavlov’s col- playing with it. This method is similar to the
laborators, extended this research to human procedure of counterconditioning described
Classical Conditioning 13

previously and constitutes the bases for the devel- a way to counteract the anxiety; (c) Exposure to
opment of systematic desensitization. the anxiety-related stimuli: While in the relaxed
Systematic desensitization: Pavlov’s and state, the patient is gradually exposed to the
Watson’s research showed that problematic anxiety-eliciting items beginning with the object
behaviors such as fear could be explained by with the lowest level in the scale developed in
classical conditioning. Jones’ experiment with Step 1 until the patient reports no fear. Afterwards,
Peter also showed that fear could be eliminated the patient is exposed progressively to the next
by gradual exposure to the fear-eliciting stimuli. items in the anxiety scale. Wolpe’s initial reports
This last result was further developed and system- were followed by numerous publications showing
atized by the work of Joseph Wolpe. Wolpe, a the effectiveness of SD on a variety of disorders.
South African psychiatrist, developed a model Exposure therapy: Although Wolpe’s work is
for the acquisition and treatment of fear and anx- commonly described as one of the beginnings of
iety that was based on his previous work with cats exposure therapy, the basic principles are different
as subjects. In his experimental work, cats as systematic desensitization is based on the coun-
received pairings of an auditory cue with a foot terconditioning procedure while exposure therapy
shock in an experimental chamber. After a few is based on extinction, and as such does not nec-
pairings, cats acquired fear to the auditory stimu- essarily involve the previous induction of an
lus, which also generalized to the experimental incompatible state. Exposure therapy is a type of
cage. According to Wolpe’s report, mere exposure cognitive-behavioral therapy in which the patient
(i.e., extinction) to the experimental cage failed in is confronted with the stimulus or event that is
reducing fear responses, and thus he evaluated associated with the problematic behavior (hence
whether a response contrary to fear was more the name). For example, an exposure-based
effective in decreasing it than exposure alone. approach for a phobia to spiders will involve
Cats were gradually exposed and fed in situations exposing the patient, gradually or otherwise, to
more and more similar to the fear-inducing cage pictures of or real spiders until the phobic symp-
until the animals managed to eat in the experimen- toms recede.
tal cage. A similar approach was used to eliminate Exposure treatments have several variations
fear to the auditory cue. depending on the procedure. On the one hand,
Based on these findings, Wolpe suggested the exposure can be graded as in systematic desensi-
same manipulation could be used to reduce fear tization or can begin with the stimulus with the
and anxiety in human patients. Although the pro- highest anxiety level, in a procedure known as
cedure was first developed by Watson y Jones, it flooding. Exposure itself can be conducted
was Wolpe who systematized and published the in vivo, that is, in direct presence of the problem-
steps of what he termed “systematic desensitiza- atic object. For example, a patient with social
tion” (henceforth SD; Wolpe 1961). Wolpe based anxiety might be instructed to give a speech to
the treatment on what he called “reciprocal inhi- an audience. Imaginal exposure, on the other
bition,” the idea that anxiety could be “inhibited” hand, involves instructing the patient to imagine
by inducing an incompatible emotional state. As vividly the object or situation. This is particularly
designed by Wolpe, a typical SD procedure has useful when the anxiety-eliciting events are not
three steps: available as, for example, with patients with post-
(a) Establish a hierarchy of anxiety-related traumatic stress disorder. In the last years, virtual
stimuli: The therapist and the patient identify the reality exposure has become more common with
objects or stimuli that are associated with anxiety. the development of technology.
Each item is given a rating based on the level of Research has shown that cognitive-behavioral
anxiety it provokes; (b) Train an incompatible therapy, and especially exposure therapy, is effec-
response: The most used incompatible response tive in alleviating the symptomatology of many
is relaxation training such as meditation or breath- disorders, including but not limited to phobias,
ing exercises. Patient are taught these strategies as
14 Classical Conditioning

PTSD, OCD, and reactivity to drug and food- animals and humans that deepened extinction suc-
related cues. cessfully prevents relapse caused by the passage
Relapse and prevention of relapse: As men- of time (spontaneous recovery) and by the presen-
tioned before, the effect of experimental extinc- tation of an aversive stimulus (reinstatement).
tion is frequently temporary, and the extinguished
response can recover. In the same manner, expo-
sure therapy does not destroy the potential of the Current Status of the Study of Classical
phobic object to cause the problematic response. Conditioning
Thus, the circumstances under which the
extinguished response recovers have served as a The principles of classical conditioning are related
model for the study of relapse in exposure therapy to many areas of Psychology, but their study by
and its prevention. Relapse occurs frequently neuroscientist has been prolific in continuously
when the patient has experience with the aversive showing the potential neural correlates to different
event outside of the treatment context (renewal), phenomena of classical conditioning.
and when some time has passed since the last Several research suggested relationships
exposure session (spontaneous recovery). between phenomena previously described here
Basic research has showed that some manipu- like blocking, conditioned inhibition, and extinc-
lations have potential to prevent relapse. One of tion and the activity of the brain. Furthermore,
them is massive extinction. In animal studies, it neuroscience also has shown some evidence that
refers to the amount of extinction trials received relates the activity of the brain with specific theo-
by the subjects, which can go up to ten times the ries of conditioning. One example of this evidence
number of trials received by a control group. The is the relationship between dopaminergic and
results have showed that massive exposure is suc- norepinephrinergic activity and error correction.
cessful in attenuating and even completely pre- Structures like the basal ganglia, the cerebellum,
venting relapse, although so far it has only been and striatum among others appear to be involved
examined in specific experimental preparations. in predicting rewards and subsequent behavior
A second treatment is conducting exposure in (Schultz and Dickinson 2000).
many contexts as opposed to in one. Several stud- Many neuroscientists that work on areas
ies have reported in different experimental prepa- related to the neural bases of learning choose to
rations that exposure in multiple contexts work on classical conditioning because it has been
attenuates relapse, although its effectiveness extensively researched and it provides with sim-
might be qualified by several factors, such as the ple yet useful learning situations to analyze.
number of contexts in which fear was acquired, or Likewise, classical conditioning is closely
the similarity between the exposure and/or acqui- related to cognitive Psychology. As suggested
sition contexts. Another technique is the use of above, many theorists of learning have stated
extinction cues or retrieval cues for extinction. An that perception, attention, and other basic pro-
extinction cue is a stimulus that is presented dur- cesses affect how we learn about the environment.
ing exposure and then is used after it is over. Furthermore, a particular approach to cognitive
Extinction cues have been shown to attenuate processes called “connectionism” often uses a
relapse in both animal and human studies, and principle called “delta rule” that closely resembles
with different experimental preparations. Finally, Rescorla-Wagner’s rule for error correction.
a less investigated treatment is what in the animal These models understand various cognitive phe-
studies is called “deepened extinction.” In this nomena through multiple parallel associations
manipulation, the patient is exposed simulta- between inputs (stimulus) and outputs
neously to multiple anxiety-eliciting cues. This (response), which bear many similarities with the
is supposed to produce a better extinction (i.e., principles of classical conditioning.
reduction in reactivity to cues) than exposure to a All these links between classical conditioning
single cue. Some studies have shown in both and other areas of research, together with the
Classical Conditioning 15

continuously growing clinical application that References

have come from the study of classical condition-
ing, show that this is still a very active area. Bouton, M. E. (1993). Context, time, and memory retrieval
in the interference paradigms of Pavlovian learning.
Moreover, the theories of the mechanism behind
Psychological Bulletin, 114(1), 80–99.
classical conditioning are still being updated and Bouton, M. E., & Bolles, R. C. (1979). Contextual control
becoming more and more complex in a of the extinction of conditioned fear. Learning and
continuous rate. Motivation, 10, 445–466.
Brown, P. L., & Jenkins, H. M. (1968). Auto-shaping of the
pigeon’s key-peck. Journal of the Experimental Anal-
ysis of Behavior, 11(1), 1–8.
Delamater, A. R., & Oakeshott, S. (2007). Learning about
Cross-Reference multiple attributes of reward in Pavlovian conditioning.
Annals of the New York Academy of Sciences, 1104(1),
▶ Aaron Blaisdell 1–20.
▶ Associative Learning Dickinson, A., & Shanks, D. (1985). Animal conditioning
and human causality judgment. In L.-G. Nilsson &
▶ Avoidance T. Archer (Eds.), Series in comparative cognition and
▶ B F Skinner neuroscience. Perspectives on learning and memory
▶ Blocking (pp. 167–191). Hillsdale: Lawrence Erlbaum Associ-
▶ Conditioned Inhibition ates, Inc.
Domjan, M. (2005). Pavlovian conditioning: A functional
▶ Conditioned Place Preference perspective. Annual Review of Psychology, 56,
▶ Conditioned Response 179–206.
▶ Conditioned Stimulus Gallistel, C. R., & Gibbon, J. (2000). Time, rate, and
▶ Contiguity conditioning. Psychological Review, 107(2), 289–344.
Garcia, J., & Koelling, R. A. (1966). Relation of cue to
▶ Contingency consequence in avoidance learning. Psychonomic Sci-
▶ Counterconditioning ence, 4, 123–124.
▶ David Hume Gormezano, I. (1966). Classical conditioning. In J. B.
▶ Disgust Sidowski (Ed.), Experimental methods and instrumen-
tation in psychology (pp. 385–420). New York:
▶ Extinction in Learning McGraw-Hill.
▶ Higher Order Conditioning Hollis, K. L. (1997). Contemporary research on Pavlovian
▶ Inhibition of Delay conditioning: A “new” functional analysis. American
▶ Ivan Pavlov Psychologist, 52, 956–965.
Jones, M. C. (1924). The elimination of children’s fears.
▶ J Bruce Overmier Journal of Experimental Psychology, 7, 382–390.
▶ John B. Watson Kamin, L. J. (1969). Predictability, surprise, attention, and
▶ Karen Hollis conditioning. In B. A. Campbell & R. M. Church
▶ Latent Inhibition (Eds.), Punishment and aversive behavior. New York:
Appleton-Century-Crofts.
▶ Learning Lubow, R. E., & Moore, A. U. (1959). Latent inhibition:
▶ Michael Domjan The effect of nonreinforced pre-exposure to the condi-
▶ Operant Conditioning tional stimulus. Journal of Comparative and Physio-
▶ Overshadowing logical Psychology, 52(4), 415–419.
Mackintosh, N. J. (1975). A theory of attention: Variations
▶ Ralph Miller in the associability of stimuli with reinforcement. Psy-
▶ Reflex chological Review, 82(4), 276–298.
▶ Release from Inhibition Miguez, G., Laborda, M. A., & Miller, R. R. (2014).
▶ Skinner Box Classical conditioning and pain: Conditioned analgesia
and hyperalgesia. Acta Psychologica, 145, 10–20.
▶ SOP Model Miller, R. R., & Matzel, L. D. (1988). The comparator
▶ Spontaneous Recovery hypothesis: A response rule for the expression of asso-
▶ Taste Aversions ciations. In G. H. Bower (Ed.), The psychology of
▶ Tolerance learning and motivation: Advances in research and
theory (Vol. 22, pp. 51–92). San Diego: Academic.
▶ Unconditioned Response Pavlov, I. P. (1927). Conditioned reflexes (G.V. Anrep,
▶ Unconditioned Stimulus Ed. & Trans.). London: Oxford University Press.
16 Classical Conditioning

Pearce, J. M. (1987). A model for stimulus generalization & W. F. Prokasy (Eds.), Classical conditioning II:
in Pavlovian conditioning. Psychological Review, 94, Current theory and research (pp. 64–99). New York:
61–73. Appleton-Century Crofts.
Pearce, J. M., & Hall, G. (1980). A model for Pavlovian Rusiniak, K. W., Hankins, W. G., Garcia, J., & Brett, L. P.
learning: Variations in the effectiveness of conditioned (1979). Flavor-illness aversions: Potentiation of odor
but not of unconditioned stimuli. Psychological by taste in rats. Behavioral and Neural Biology, 25(1),
Review, 87(6), 532. 1–17.
Pearce, J. M., & Mackintosh, N. J. (2010). Two theories of Schultz, W., & Dickinson, A. (2000). Neuronal coding of
attention: A review and a possible integration. In prediction errors. Annual Review of Neuroscience,
M. Chris & M. E. Le Pelley (Eds.), Attention and 23(1), 473–500.
associative learning: From brain to behavior Siegel, S., Baptista, M. A., Kim, J. A., McDonald, R. V., &
(pp. 11–39). Oxford, UK: Oxford University Press. Weise-Kelly, L. (2000). Pavlovian psychopharmacol-
Rescorla, R. A. (1969). Pavlovian conditioned inhibition. ogy: The associative basis of tolerance. Experimental
Psychological Bulletin, 72(2), 77–94. and Clinical Psychopharmacology, 8(3), 276–293.
Rescorla, R. A. (1988). Pavlovian conditioning: It’s not Wagner, A. R. (1981). SOP: A model of automatic memory
what you think it is. American Psychologist, 43(3), processing in animal behavior. In N. E. Spear & R. R.
151–160. Miller (Eds.), Information processing in animals:
Rescorla, R. A., & Heth, C. D. (1975). Reinstatement of Mem- ory mechanisms (pp. 5–47). Hillsdale: Erlbaum.
fear to an extinguished conditioned stimulus. Journal Watson, J. B., & Rayner, R. (1920). Conditioned emotional
of Experimental Psychology: Animal Behavior Pro- reactions. Journal of Experimental Psychology, 3,
cesses, 1, 88–96. 1–14.
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Wolpe, J. (1961). The systematic desensitization treatment
Pavlovian conditioning: Variations in the effectiveness of neuroses. Journal of Nervous and Mental Disease,
of reinforcement and nonreinforcement. In A. H. Black 132, 189–203.