ArticlePDF Available

Figures

Content may be subject to copyright.
International Journal of Comparative Psychology, 2004, 17, 279-303.
Copyright 2004 by the International Society for Comparative Psychology
A Review of the Empirical Laws of Basic Learning
in Pavlovian Conditioning
Martha Escobar
Auburn University, U.S.A.
and
Ralph R. Miller
State University of New York at Binghamton, U.S.A.
Contemporary learning research has provided multiple paradigms that have benefited not only researchers in
the field, but also applied theorists and practitioners. However, the emphasis on theory development has
made the learning literature almost impenetrable to nonexperts. In the present paper, we attempt to summa-
rize not the different theoretical perspectives that have been proposed to explain different instances of learn-
ing, but the empirical relationships that testing of such theories has uncovered. Because the empirical rela-
tionships we summarize here hold across preparations and species, we suggest that such relationships should
be understood as the empirical laws of basic learning. The focus of our review is the Pavlovian conditioning
tradition, but most of these relationships also apply to instrumental learning and causality learning. We hope
that the relatively novel organization we present here helps researchers and practitioners to directly incorpo-
rate these empirical principles into their current theoretical framework, whatever it may be.
The study of behavioral change can be traced as far back as civilization itself.
The works of Aristotle, Plato, Kant, and Hume (among others) reflect the philosophical
roots of the study of behavior change. However, the rigorous scientific study of behav-
ioral change started only about 120 years ago with the work of Ebbinghaus (1885/1913),
Thorndike (1898), and Pavlov (1927), among others. These early scientific endeavors
were largely driven by empirical phenomena, although theorizing was also evident. For
example, Pavlov suggested that different stimuli were represented as “centers” in the
brain, which were linked either by nature (as in the case of an unconditioned stimulus,
US, and an unconditioned response, UR) or by experience (as in the case of a conditioned
stimulus, CS, and a US). Similarly, Thorndike's (1911) Strong Law of Effect posited that
there is no learning without reinforcement because reinforcers are required to engage the
learning mechanism.
A problem with these theoretical interpretations was that many researchers
viewed Pavlov’s and Thorndike’s theoretical accounts as definitions of learning. Thus,
Pavlov’s stimulus substitution view cemented the belief that, in classical conditioning,
learning occurs only if the US can elicit a response without learning, overlooking phe-
nomena such as second-order conditioning (e.g., Pavlov, 1927) and sensory precondition-
ing (Brogden, 1939; see below). Similarly, Thorndike’s (1911) Strong Law of Effect ce-
mented the belief that no learning could occur in the absence of reinforcement, which
obviates situations in which learning occurs without apparent shifted reinforce-
ment (e.g., song learning by birds). Thus, on many occasions, the focus from the
Support for its preparation was provided by the National Institute of Mental Health (USA) Grant 33881. The
authors would like to thank Jeffrey Amundson, Francisco Arcediano, Raymond Chang, Elaina Frieda,
Randolph Grace, Matthew Johnson, Scott Parker, Steven Stout, and Dan Wheeler for their constructive com-
men ts. Correspondence sh ould be addressed to R. R. Miller, Department of Psychology, SUNY-Binghamton,
Binghamton, NY 13902-6000, U.S.A. (rmiller@binghamton.edu).
-280-
empirical phenomenon to the theoretical explanation of it. In part as a reaction to this
problem, Skinner (e.g., 1938, 1950) stressed the importance of identifying empirical rela-
tionships without constraining the interpretation of the data by a preexisting theory. We
view both the theory-bound and the antitheoretical perspectives as extremes, and we opt
for an intermediate position. Theoretical developments are essential for the advancement
of psychological science. Without theory, research would have no direction and there
would be no opportunity to integrate disparate observations into a common framework
(Platt, 1964; Popper, 1959). However, it is critical that we not confuse theory with em-
pirical observations. Psychological theories come and go, but the empirical relationships
that they must explain remain. That is, we must distinguish between empirical phenom-
ena (i.e., data observed in the laboratory) and theoretical accounts of what has been ob-
served (i.e., attempts to explain the empirical observations).
Here, we attempt to summarize the empirical relationships that appear to be ro-
bust and generalize most widely across tasks and species, avoiding theoretical interpreta-
tions where possible. With this goal in mind, phenomena are organized according to simi-
larity of treatments and their effects on behavior rather than on presumed underlying
mechanisms. In the present paper, learning will not refer to the [cognitive or neural]
processes that underlie behavioral change, but rather to the behavioral change that is ob-
served as a result of the training experience. We will draw our laws of learning mostly
from the classical (i.e., Pavlovian) learning literature, which refers to behavioral change
induced by contingencies between cues and outcomes. However, most of what we have to
say readily translates into instrumental and causality learning situations (see below).
As a final note, the review presented here is intended to be informative, not ex-
haustive. Readers interested in a more in-depth review of the empirical laws that we de-
scribe here can find such a review in Miller and Escobar (2002).
Stimulus Salience
Salience is often used to refer to a variable that depends upon both the state of
the subject and the physical characteristics of the stimulus (i.e., cue or outcome) in ques-
tion. Here we concentrate on how stimulus salience, understood exclusively in terms of
the attributes of the stimulus, affects the acquisition of behavior. The salience of a cue
generally is positively correlated with how fast conditioned responding is acquired (i.e., it
influences the rate of acquisition; Kamin, 1965), and the salience of the outcome is gen-
erally positively correlated with how much conditioned responding is observed (i.e., it
influences the asymptotic magnitude or probability of conditioned responding; e.g.,
Kimble, 1955). Simply stated, more salient stimuli promote faster learning and more
conditioned responding than less salient stimuli. The best known determinant of stimulus
salience is surely intensity, but other determinants of stimulus salience include stimulus
size, motion, contrast, and stimulus change, among others. Importantly, stimulus salience
has facilitative impact on behavior not only during training, but also during testing (cf.
Hull, 1952; e.g., Kamin, 1965).
-281-
Similarity and Contiguity
Some of the 18th century British empiricists (e.g., Berkeley, 1710/1946) cor-
rectly recognized the importance to learning of stimulus similarity. If the cue and out-
come share some attributes, learning will occur faster than if they do not share any attrib-
utes (e.g., Rescorla & Furrow, 1977). For example, two lights or two tones are more
readily associated than a light and a tone; however, it is unclear whether this facilitatory
effect arises from stimulus generalization between the cue and outcome and/or enhance-
ment of the effects of the pairings. Stimulus similarity is defined not only with respect to
the what aspects of the stimuli (i.e., modality, size, shape, texture, etc.), but also with
respect to their where (space) and when (time). Thus, contiguity in space and time might
be viewed as special instances of stimulus similarity. Despite there being some compel-
ling demonstrations of the importance of stimulus similarity for associative learning (e.g.,
Krane & Wagner, 1975; Rescorla & Furrow, 1977; Testa, 1975), the importance of
stimulus similarity for learning in dimensions other than time and space has not been suf-
ficiently emphasized in recent years.
In general terms, contiguity between the cue and the outcome favor the develop-
ment of conditioned responding. The effects of contiguity are observed in both the tem-
poral domain (Pavlov, 1927) and the spatial domain (Rescorla & Cunningham, 1979).
Temporal contiguity has received the most attention in learning research, with some theo-
rists proposing that it should be considered a sufficient condition for learning to occur
(e.g., Estes, 1950; Guthrie, 1935). However, despite contiguity being maximal when the
cue and outcome occur simultaneously, the usual observation is that behavioral control
by the cue-outcome relationship appears to be stronger if the cue immediately precedes
the outcome during training, or even if a small interval separates termination of the cue
from the onset of the outcome, than if they are simultaneous. We will address this simul-
taneous conditioning deficit below.
One must note that the effects of contiguity vary depending on the stimuli and
preparations used to assess the acquisition of new behavior. For example, in eyelid condi-
tioning, conditioned responding occurs only when the interval between onset of the cue
and onset of the outcome is less than 1 s (e.g., Kimble, 1947; McAllister, 1953). In con-
trast, in conditioned taste aversion training, conditioned responding occurs even if several
hours separate the taste and toxin (e.g., Garcia, Ervin, & Koelling, 1966). Despite these
differences, in all cases the relationship between cue-outcome intervals and amount of
conditioned responding follows the same function: As the interval increases from zero,
conditioned responding briefly increases until a most effective interval is reached, and
then it smoothly decreases until a complete absence of responding is reached.
Sufficiency of Contiguity
The sufficiency of contiguity for the development associative learning requires
little consideration because of the many situations in which learning does not occur de-
spite the cue and outcome being contiguous. For example, if the cue is of very low sali-
ence, learning will be weak despite strong contiguity. Similarly, variations in cue-
outcome contingency (discussed below) can override the effects of contiguity. Like high
stimulus salience, all of the factors on this list are included because they were found to
influence learning; thus, contiguity by itself is in no sense of the word “sufficient.”
-282-
Necessity of Contiguity and Mediation
Conditioned responding to a target stimulus, X, not only reflects the associative
history of X. Instead, it reflects the associative history of both X and other cues associ-
ated with X. That is, conditioned responding to X can be mediated by another stimulus,
M. Pavlov (1927) observed that a stimulus that elicits a conditioned response could be in
turn used to train a conditioned response to another stimulus. Thus, in second-order
conditioning, mediating Stimulus M is paired with the outcome (M-outcome) with the
usual observation that M comes to elicit a conditioned response. Subsequently, the target
cue, X, is paired with M (i.e., X-M), which acts as a surrogate US. This type of training
often results in conditioned responding to Cue X. A closely related phenomenon is sen-
sory preconditioning (Brogden, 1959). In sensory preconditioning, the mediating stimu-
lus, M, and the target cue, X, are paired before M is paired with the outcome (i.e., M-
outcome). This treatment often results in conditioned responding to X. Second-order
conditioning and sensory preconditioning differ principally in the order in which the X-M
and M-outcome pairings are given to the subject (M-outcome then X-M in second-order
conditioning, and X-M then M-outcome in sensory preconditioning). The same type of
mediated learning (i.e., conditioned responding to Cue X even when it has not been
paired with the outcome) can also be produced when the X-M and M-outcome training
trials are interspersed (e.g., Rashotte, 1981; Yin, Barnet, & Miller, 1994).
Complicating the phenomenon of mediated learning is that there appears to be
two opposing types of mediation, which we shall call positive and negative. The previ-
ously mentioned phenomena of sensory preconditioning and second-order conditioning
are two examples of positive mediated responding, in which conditioned responding to
the target (X) varies directly with responding to the mediating stimulus (M). In contrast,
in negative mediated responding, conditioned responding to the target (X) varies in-
versely with responding to the mediating stimulus (M). That is, if a manipulation in-
creases (or decreases) conditioned responding to the mediating stimulus, responding to
the target decreases (or increases). The best known example of negative mediation is
conditioned inhibition (for reviews, see LoLordo & Fairless, 1981; Rescorla, 1969; Sa-
vastano et al., 1999), which might be viewed as the learning of a relationship between a
cue and the omission of the outcome. There are many training procedures that result in
the development of conditioned inhibition, but perhaps the best known is Pavlov’s
(1927) procedure, in which pairings of the target cue (X) and mediating (M) stimuli (i.e.,
X-M) are interspersed with pairings of the mediating stimulus and the outcome (M-
outcome). After this training, subjects presented with X behave as if X predicts the omis-
sion of the outcome (e.g., subjects withdraw from a conditioned inhibitor for food;
Wasserman, Franklin, & Hearst, 1974).
Note the similarity between the procedures used to obtain positive and negative
mediation: In both cases, there are X-M pairings and M-outcome pairings. Recent re-
search suggests that that the number of pairings of the target cue and mediating stimulus
(i.e., X-M pairings) is at least one critical factor in determining whether positive or nega-
tive mediation will be observed. In general, few target cue-mediating stimulus pairings
results in behavior consistent with positive mediation (i.e., second-order conditioning),
but the behavior becomes consistent with negative mediation as the number of pairings
increases (i.e., conditioned inhibition; e.g., Rashotte, 1981; Yin et al., 1994).
-283-
Relativity of Contiguity
The influence on learning of cue-outcome temporal contiguity is not absolute.
Rather, it appears to be relative to the interval between successive outcomes (i.e., inter-
trial interval; see Gibbon et al., 1977; Gibbon & Balsam, 1981). That is, although the
long-term effects of training appear to be favored by short cue-outcome intervals, they
appear to be hampered by short intervals between successive outcomes. This deleterious
effect is called the trial-spacing effect: Massed cue-outcome pairings result in the rapid
development of acquired responding (i.e., fewer trials are required to reach asymptote)
which is then largely lost over long retention intervals, whereas spaced cue-outcome pair-
ings result in the slower development of acquired responding (i.e., more trials are re-
quired to reach asymptote) which is better maintained over long retention intervals (for
examples with humans, see Cook 1934; Hovland, 1940). The implications for education
and skill training are as follows: Learned materials will be available to the subject for
longer periods of time if practice of such material is distributed over time. In contrast,
fast learning of material that need not be remembered over long periods of time will
benefit from massed practice. Clearly, the desirability of spaced training trials relative to
massed ones depends upon one’s goals in terms of number of training trials, temporal
duration of training, and retention interval. But, what if the goal is to acquire material
quickly and maintain it for a long time? Landauer and Bjork (1978; also see Bjork, 1988)
demonstrated that the best strategy to insure fast learning and long-term retention is to
combine the two forms of trial spacing. Landauer and Bjork asked subjects to imagine
that they were at a cocktail party and they were to learn the names of several people they
met at that party. The names were presented in a massed schedule, a spaced schedule, or
an expanding schedule (in the latter schedule, the intertrial interval increased with re-
peated trials). Their results suggested that the expanding schedule was the optimal strat-
egy: Seemingly, massing in the first few trials enhanced recall in the immediate subse-
quent trials, and spacing in later trials enhanced long-term recall of the learned material.
Decomposing Stimuli as a Function of Time
There is abundant evidence that subjects perceive a stimulus as a series of ele-
ments, using not only stimulus onset as a cue for behavior, but also stimulus presence
(often with respect to time since onset; e.g., Romaniuk & Williams, 2000) and stimulus
termination. Yet, for all stimulus components (e.g., stimulus onset, each instant of stimu-
lus presentation, and stimulus termination), proximity to the outcome appears to deter-
mine the degree to which the learned behavior will be exhibited. For example, animals
trained with a long auditory signal of shock will not exhibit fear-related behaviors until
the last few seconds of presentation of the cue (i.e., inhibition of delay, cf. Pavlov, 1927;
Rescorla, 1967). Thus, decomposition of stimuli into component parts complicates
analysis, but does not appear to compromise the importance of contiguity for the occur-
rence of learning.
-284-
Spatial Contiguity
Today we recognize that spatial contiguity, as well as temporal contiguity, is an
important determinant of learning. All other things being equal, cues and outcomes that
occur close in space become better associated than cues and outcomes that occur far apart
from each other. For example, Rescorla and Cunningham (1979) trained pigeons with a
pair of events. With identical temporal relationships, in one group the two events oc-
curred in the same spatial location, whereas in the other group the two events occurred in
different spatial locations. Although pigeons in both groups acquired an association be-
tween the two events, the group for which the two events occurred in the same spatial
location acquired the association faster than the group for which the two events occurred
in different spatial locations. Thus, contiguity can be characterized as a spatiotemporal
variable which impacts learning across almost all situations and parameters.
Predictive Value as an Alternative to Contiguity
We have already discussed the problem that mediated learning poses for contigu-
ity. An even greater problem is the observation that more robust behavior control appears
to occur when a cue slightly precedes an outcome (i.e., delay conditioning) than when the
cue and outcome are simultaneously presented during training (simultaneity represents
maximal contiguity). This simultaneous conditioning deficit suggests that contiguity
may not be a necessary condition for learning to occur; indeed, it may even hamper learn-
ing. However, one must realize that most experimental situations require subjects to “an-
ticipate” the outcome, and it would be functionally inappropriate to respond to a simulta-
neous cue which effectively announces that the outcome is present “now” as if it an-
nounced that “it is coming” (e.g., Matzel, Held, & Miller, 1988; Savastano & Miller,
1998). Esmoris-Arranz, Pardo-Vázquez, and Vázquez-García (2003) observed that rats
freeze but fail to exhibit a flight response when trained with a delayed signal-shock rela-
tionship, and exhibit flight but not freezing responses when trained with a simultaneous
signal-shock relationship. This observation is consistent with rats’ response to danger in
the natural environment: Rats freeze when presented with cues that allow them to antici-
pate immediate danger, but vocalize and take flight when presented with cues that indi-
cate danger in the current situation. That is, the specific form (i.e., topology) of the condi-
tioned response changes with the temporal relationship between the cue and the outcome
(e.g., Burns & Domjan, 1996; Timberlake & Lucas, 1991). Most conditioned fear prepa-
rations use freezing measures, which require rats to anticipate, rather than deal with im-
mediate danger.
Importantly, the order in which the subject experiences the paired events also in-
fluences the behavior that is ultimately observed. For example, forward pairings
(cueoutcome) usually result in excitatory conditioned responding to the cue, whereas
backward pairings (outcomecue) of the same cue and outcome initially result in weak
excitatory conditioned responding to the cue, which changes to behavior consistent with
conditioned inhibition with subsequent trials (e.g., Heth, 1976). Regardless of the order
of the cues, neither excitatory nor inhibitory behavioral control is observed when the
temporal separation of the cue and outcome exceeds a certain threshold duration (e.g.,
Miller et al., 1991).
-285-
Predispositions
People and other animals never enter a learning situation tabulae rasae. They ar-
rive with predispositions that result in more rapid learning of some cue-outcome dyads
than others. Probably the most cited example of such predispositions is Garcia and Koel-
ling’s (1966) “bright-noisy water” experiment. In this study, Garcia and Koelling found
that rats acquired flavor aversion more rapidly when the flavor was paired with an inter-
nal malaise (induced upset stomach) than when it was paired with footshock, whereas
they acquired fear of an audiovisual cue more rapidly when it was paired with the foot-
shock than when it was paired with the internal malaise. That is, rats are more predis-
posed to associate ingested flavors with internal stimuli (such as stomach upset) than
lights and sounds, whereas they are more predisposed to associate external painful stim-
uli (such as footshock) with lights and sounds than ingested flavors. Predispositions are
sometimes called cue-to-consequence effects, and were anticipated by Thorndike’s
(1932) concept of “belongingness.”
Garcia originally viewed predispositions as genetic in origin, specifically a result
of natural selection in the animal’s ecological niche. According to this view, animals are
more prepared to learn about stimulus dyads that are functional in their natural habitat
than about nonfunctional dyads. Thus, rats, which in their natural habitat find food pri-
marily through olfactory and gustatory cues, are more predisposed to associate internal
malaise with odors and tastes than with audiovisual cues. In contrast, rodents avoid po-
tential predators mostly by using auditory cues, which is consistent with the observation
that audiovisual stimulus-shock associations were favored. Similarly, humans exposed to
chemotherapy or radiotherapy tend to report conditioned nausea when presented with
cues associated to the chemical or radiological treatment (e.g., Stockhorst, Klosterhalfen,
& Steingrueber, 1998). Moreover, conditioned nausea occurs more readily to smells and
tastes than to visual cues (e.g., Cameron et al., 2001). Thus, humans too are more predis-
posed to associate smells and tastes than visual cues with gastrointestinal upset. How-
ever, research has made clear that prior experience interacts with genetic predispositions
to determine which stimulus dyads will be favored in learning. For example, Dalrymple
and Galef (1981) found that, although rats are initially slow to associate visual cues and
illness, they become better at doing so after extended experience with these cue-outcome
pairs (consistent with these studies, chemotherapy patients that have received extensive
treatment report nausea when approaching the hospital or reading hospital signs in the
highway). Thus, like all other aspects of behavior, predispositions to establish stimulus
control of behavior appear to reflect an interaction of genes and prior experience.
Predispositions appear to be so embedded in the genetic makeup of subjects that
learning consistent with these predispositions occurs extremely rapidly, and might even
occur as a result of observing another individual interacting with the cues and outcomes.
For example, Mineka and her colleagues (e.g., Mineka et al., 1984) have demonstrated
that monkeys acquire conditioned fear of plastic snakes much faster than they acquire
conditioned fear of plastic flowers (i.e., they are predisposed to fear snakes but not flow-
ers). Interestingly, this difference is also observed if subjects do not directly interact with
the snake or flowers, but instead merely observe a video of another monkey interacting
with the snake or flower (Cook & Mineka, 1990).
By and large, humans and other species are predisposed such that most specific
instances of learning are functional (i.e., they appear to serve the goals of survival and
-286-
genetic propagation). An excellent example is the influence of learning on sexual behav-
ior. Hollis and her colleagues (e.g., Hollis, Cadieux, & Colbert, 1989) have demonstrated
that Pavlovian signals of female accessibility decrease aggressive behavior in territorial
fish, which in turn increases the likelihood of a successful encounter between the male
and the female (see Hollis, 1997, for other examples of functional analyses of Pavlovian
behavior). However, there are numerous documented examples in which specific in-
stances of learned behavior are not functional, but instead detrimental to the well being of
an organism. Typically, these instances arise in situations that create contingencies con-
trary to those that prevail in the animal's natural habitat or are inconsistent with its past
experience. One type of dysfunctional acquired responding is illustrated in vicious circle
behavior (Gwinn, 1949). Typically, in Phase 1, rats are placed in a start box and the be-
ginning of the trial is signaled by the experimenter’s lifting a barrier and delivering foot
shock in the start box. Rats must then run through an electrified runway to reach a safe
goal box. In Phase 2, the start box is not electrified anymore, but the animals run through
the electrified runway to reach the goal box anyway, even though they could avoid shock
altogether by staying in the start box. That is, their learned behavior prevents them from
experiencing the new contingency (i.e., safety in the start box). A second type of dysfunc-
tional behavior is illustrated in negative automaintenance (Williams & Williams, 1969),
in which a Pavlovian conditioned response (pecking by pigeons at a cue followed by
food) causes the omission of reward (food delivery), with the result that the subject keeps
responding (albeit at a reduced rate) despite its pecking causing a decrease in reinforce-
ment. This observation is not surprising if one considers that for pigeons, over many gen-
erations, pecking has been necessary to obtain food and omission of food for pecking is
inconsistent with contingencies in their natural habitat.
Table 1.
2 x 2 Contingency Table for a Single Cue and Single Outcome
Outcome present Outcome absent
Cue present Cell a
# trials with cue and outcome paired
Cell b
# trials with cue alone
Cue absent
Cell c
# trials with outcome alone
Cell d
# trials with cue and outcome absent
Contingency
Contingency is a term meant to convey the correlation between the cue and out-
come. Contingency increases as the number of trials on which the cue and outcome are
presented or omitted together increases, and it decreases as the number of trials on which
either the cue or outcome are presented alone increases. Table 1 presents a contingency
table with the four possible types of events in a learning situation: Cue-outcome pairings
(Cell a), cue alone (Cell b), outcome alone (Cell c), and absence of both the cue and out-
come (Cell d). In contrast with contiguity, which speaks to the quality of the cue-outcome
pairings (i.e., cue-outcome spatiotemporal proximity), contingency speaks to the reliabil-
ity of these pairings (independent of their contiguity). Many different algebraic formulas
have been proposed to represent the impact of contingency on behavior, and each of these
formulas makes assumptions about the relative values (weights) of each of the four types
-287-
of events summarized in Table 1. Research so far has failed to identify any one measure
that, across tasks and parameters, consistently and accurately describes the resultant
stimulus control. However, there is little argument that learning is facilitated by events
that confirm a cue-outcome relationship (i.e., when the cue and outcome are paired and
when neither event occurs; Cells a and of Table 1) and decremented by events that deny a
cue-outcome relationship (i.e., occurrences of either the cue or outcome in the absence of
the other; Cells b and c of Table 1). Learning of a cue-outcome association benefits the
most from a-type (i.e., cue-outcome) trials, the least from d-type (no cue-no outcome)
trials, and is intermediately impaired by b- (cue alone) and c-type (outcome alone) trials
(e.g., Kao & Wasserman, 1993).
Table 2.
Attenuation of Stimulus Control by Degrading the Cue-Outcome Contingency as a Result of Adding Cue-
alone or Outcome-alone Presentations
Cue Outcome
Before cue-outcome pairings Latent inhibition US-preexposure effect
Interspersed during cue-outcome
pairings Partial reinforcement Degraded contingency effect
After cue-outcome pairings Extinction US-postexposure effect
Note. Training here refers to cue-outcome pairings. US = an unconditioned stimulus serving as an out-
come; US is used here because this procedure for degrading stimulus control has traditionally been stud-
ied in Pavlovian situations.
Degrading a contingency refers to adding presentations of the cue alone and/or
the outcome alone to a situation that includes some cue-outcome pairings. The added
events can occur before, interspersed among, or after the pairings, creating the six possi-
ble situations presented in Table 2. First, let us consider added cue-alone presentations.
When these cue-alone presentations occur before the cue-outcome pairings, latent inhibi-
tion (a.k.a. the CS-preexposure effect) is often observed: If a cue has been repeatedly
experienced alone, subsequent pairings of that cue with an outcome are less effective in
producing a conditioned response (i.e., more pairings are required to achieve a learning
criterion; Lubow & Moore, 1959). When the cue-alone presentations occur interspersed
with cue-outcome pairings, we talk about partial reinforcement, a treatment that usually
results in slower acquisition (i.e., more CS presentations to reach asymptote) and lower
asymptotes of conditioned responding (Pavlov, 1927) as well as greater resistance to ex-
tinction of the conditioned response (e.g., Rescorla, 1999). Finally, if the cue-alone pres-
entations occur after the cue-outcome pairings, we observe extinction (i.e., gradual dissi-
pation) of the conditioned response (Pavlov, 1927).
Now, let us consider adding outcome-alone presentations. When these additional
outcomes occur prior to the cue-outcome pairings, the resultant retardation in observing
conditioned responding is called the US-preexposure effect (Randich & LoLordo, 1979).
When outcome-alone presentations occur interspersed among the cue-outcome pairings,
the resultant deficit in conditioned responding is called the degraded contingency effect
(Rescorla, 1968). Finally, when the outcome-alone presentations occur after the pairings,
the resultant deficit is sometimes called the US-posttraining exposure effect (hereafter
Temporal location of added event
Added
-288-
called the US-postexposure effect; Chang, Stout, & Miller, 2004), and it is the most dif-
ficult to observe of the six contingency degrading manipulations described in Table 2.
Some recent studies suggest that the difficulty in obtaining US-postexposure effects is
related to the so-called ‘biological significance’ acquired by the cue during the cue-
outcome training. Miller and colleagues (e.g., Miller & Matute, 1996) have reported de-
graded contingency through outcome postexposure as long as the cue and outcome are of
low biological relevance (i.e., neither elicits strong unconditioned responses prior to the
cue-outcome pairings).
The construct of contingency is challenged by the problem of defining a trial.
This problem is most evident when one tries to count the number of trials that should be
counted in Cell d of Table 1: How many no cue-no outcome trials are contained in five
minutes? How many in five seconds? The convention has been to define a trial as being
of a fixed, uniform duration with that duration being set equal to the duration of trials on
which the cue and/or outcome both occur (e.g., the duration of a typical Cell a-type trial).
Such a definition is surely arbitrary, but in most cases it suffices, as long as the same du-
ration is used to define a trial in all conditions. Of course, the entire construct of trials
exists only in the mind of the experimenter: Subjects live and process information in con-
tinuous time, not in the trial-wise manner in which time is partitioned for our research
purposes.
Primacy and Recency as Modulators of Contingency
Contingency degradation necessarily involves at least one of two different types
of trials (i.e., the cue-outcome pairings, and either cue-alone or outcome-alone presenta-
tions). In situations in which the contingency degrading trials are not interspersed among
the cue-outcome pairings (i.e., when all of one trial type precedes all of the other trial
type, namely the cases of latent inhibition, extinction, US-preexposure, and US-
postexposure), strong recency effects are observed (i.e., the resultant behavior reflects
the more recent trials). For example, subjects exposed to 100 cue-outcome trials followed
by 100 cue-alone trials will exhibit little conditioned responding, while if the training is
reversed, robust conditioned responding will be observed. These differences in condi-
tioned responding due to the order of trials during training are known as trial-order ef-
fects. Importantly, trial-order effects do not always take the form of recency effects.
Sometimes, we may observe primacy effects , which refer to greater influence on behavior
of early training experiences than later experiences. Moreover, as time elapses, recency
effects often wane allowing the expression of information favored by primacy (e.g.,
Kraemer, Randall, & Carbary, 1991). For example, after the cue-outcome pairings, the
passage of time may result in forgetting (i.e., a loss of conditioned responding) and, after
the extinction training, the passage of time may result in spontaneous recovery of the
extinguished conditioned response. Primacy effects are usually quite weak but, unlike
recency effects, they do not wane with time alone (other manipulations, such as extinction
of the context, may attenuate the effects of primacy; e.g., De la Casa & Lubow, 2002); if
anything, they grow stronger with time since the end of training (see e.g., Postman, Stark,
& Fraser, 1968; Underwood, 1948).
-289-
Context as a Modulator of Contingency.
As we previously stated, contingency degradation treatments necessarily involve
at least two types of trials: Cue-outcome pairings to establish the contingency and which
result in conditioned responding, and presentations of the cue and/or outcome alone to
degrade the contingency and which attenuate conditioned responding. If the two different
trial types occur in different contexts, behavior will tend to reflect the experience learned
in the particular context in which the subject is currently being assessed (for a review, see
Bouton, 1993)
Although we often think of context as referring to static background cues that
persist over a [training] session, contextual cues can also be more delimited in time and
space. When a discrete stimulus signals whether or not another stimulus will be followed
by reinforcement, that stimulus is refereed to as an occasion setter (Holland, 1992;
Miller & Oberling, 1998); that is, a stimulus that sets the occasion for a given contin-
gency. Research by Holland (1992) has demonstrated that an occasion setter need not
elicit the conditioned response in its own right but modulates conditioned responding to
the target stimulus. That is, the role of the occasion setter is to disambiguate the meaning
of a target stimulus rather than to produce conditioned responding itself.
Permanence of the Degraded Contingency Effects
The loss of conditioned responding observed with all of the degraded contin-
gency effects listed on Table 2 seems to be largely, if not entirely, a lapse (i.e., inaccessi-
bility of the information), as opposed to an irreversible loss, of the memory of the cue-
outcome pairings. We base this assertion on the observation that treatments other than
further cue-outcome training can result in a return of conditioned responding. As previ-
ously mentioned, extinguished conditioned responses can spontaneously recover over
long retention intervals, an observation that led Pavlov (1927) to suggest that extinction
reflected new learning rather than erasure of previous learning. Spontaneous recovery
from latent inhibition is also sometimes observed (Kraemer et al., 1991). Moreover, most
degraded contingency effects such as latent inhibition and extinction are context depend-
ent. That is, conditioned responding partially recovers if the subject is tested in a context
in which the contingency degradation did not occur (e.g., Bouton & King, 1983; Channell
& Hall, 1983). Similarly, a brief presentation of either the cue or the outcome alone often
restores conditioned responding (i.e., reminder treatments; for a review see Miller,
Kasprow, & Schachtman, 1985).
Interference
Since the 1960s, researchers concerned with basic animal learning have largely
focused on one select type of stimulus interference, specifically how cues trained together
(i.e., in compound) compete for the prediction of an outcome (i.e., cue competition). Cue
competition refers to impaired conditioned responding to a target cue that is trained in
the presence of one or more potentially competing cues. Traditionally, discussions of cue
competition refer to the overshadowing and blocking effects. The overshadowing effect
(Pavlov, 1927) refers to impaired conditioned responding to the target (overshadowed)
cue due to the presence of a (usually) more salient competing (overshadowing) cue during
-290-
training with the outcome. In the blocking effect (Kamin, 1968), the cues are usually of
equal salience, but the competing (blocking) cue receives pairings with the outcome prior
to the compound trials.
Competition between cues trained together was popularized by Kamin (1968),
who viewed it as having two important implications. First, it demonstrated that neither
contiguity nor contingency were sufficient to support learning (until that time, the suffi-
ciency of contiguity and contingency had been largely unchallenged). Take for example
the case of overshadowing. In a typical overshadowing experiment, pairing the compound
of competing Cue A and target Cue X with the outcome results in less conditioned re-
sponding to X than in a control condition in which X alone is paired with the outcome.
However, the X-outcome contiguity and contingency are the same in both groups. Sec-
ond, cue competition suggested to Kamin that, in situations with multiple cues, subjects
do not learn about each cue in isolation; rather, the cues interact with each other to con-
trol behavior. The identification of this sort of stimulus interference greatly advanced the
theoretical study of learning: Models were formulated to account for cue competition be-
tween cues that were trained together (e.g., Rescorla & Wagner, 1972). However, this
focus of the models resulted in researchers’ ignoring other forms of stimulus interference.
The 2 x 2 matrix in Figure 1 depicts four different types of stimulus interference.
In the preceding paragraph, we discussed competition between cues trained together,
which is represented by Cell 1 of Figure 1. Cell 2 is similar to Cell 1 in that it involves
stimuli trained in compound but, in this case, interference occurs between outcomes
trained in compound rather than between cues trained in compound. A potential problem
in observing competition between outcomes is that outcomes are usually of high salience
or biological relevance. Indeed, it is possible that the outcomes will be so salient that the
subject will not disregard any as being related to the cue. This is the usual case with ani-
mal subjects trained with biologically significant outcomes such as food, pain, and water.
The lack of interference with biologically significant outcomes should not be taken as
evidence of absence of interference between outcomes. If neutral stimuli are used during
training, the subject can experience both outcomes and weigh the extent to which they are
predicted without the potential distraction or relevance of biologically significant out-
comes. Indeed, Esmoris-Arranz, Matute, and Miller (1997; also see Miller & Matute,
1998; Rescorla, 1980, pp. 90-97) performed such studies with rat subjects and observed
interference between outcomes trained together. In their study, animals were trained in a
blocking preparation. Cue A (a tone) was paired with Outcome 1 (a buzzer) during the
first phase of training. Then, Cue A was paired with the compound of Outcome 1 and
Outcome 2 (a buzzer and a clicker, respectively). Cue A was then paired with a shock US
and conditioned responding to Outcome 2 was observed to be lower than in a group
which lacked the A-Outcome 1 pairings. That is, when a cue was paired with two simul-
taneous outcomes of low biological significance, its association with each outcome was
weaker than if the cue had been paired with a single outcome.
Cell 3 depicts interference between cues trained apart with the same outcome.
This paradigm was widely studied by researchers within the human verbal learning tradi-
tion in the middle of the twentieth century (for a review see e.g., Slamecka & Ceraso,
1960). In these studies, people were asked to memorize lists of word pairs, which usually
shared some common terms. For example, a first list might have contained the pair cat-
train and the second list might have contained the pair parrot-train. After memorizing the
two lists, subjects were asked which word had previously been presented together with
-291-
the common associate (in this case, the word train). These types of studies ordinarily
yielded retroactive interference (recall of the second associate, parrot) or proactive inter-
ference (recall of the first associate, cat). Cell 3-type interference is not unique to verbal
materials; it has been observed with both human participants (Matute & Pineño, 1998)
and rat subjects (Amundson, Escobar, & Miller, 2003; Escobar, Matute, & Miller, 2001;
Escobar, Arcediano, & Miller, 2001) in nonverbal preparations. For example, Amundson
et al. trained rats with pairings of auditory Cue A and a shock. Subsequently, animals
were trained with pairings of a second auditory cue, X, and the shock. They observed that
conditioned responding to X was lower than if the A-shock pairings had not occurred
(i.e., proactive interference). Thus, we see that cue competition is not limited to cues
trained in compound, but can be obtained between cues trained independently.
Cell 4 of Figure 1 describes situations in which one cue is separately paired with two dif-
ferent outcomes. The prototypical example of this type of interference is the phenomenon
of counterconditioning. In a counterconditioning paradigm, a cue is first paired with
Outcome 1 until that cue produces conditioned responding consistent with Outcome 1. In
a subsequent phase, the cue is paired with Outcome 2, and the latter learning is observed
to interfere with the original conditioned response (Pavlov, 1927; Sherrington, 1947;
Wolpe, 1958). This sort of interference has also been reported in verbal learning situa-
tions with human participants (e.g., Postman, 1962) and with outcomes that do not have
biological significance in rat subjects (Escobar, Arcediano, & Miller, 2001). Countercon-
ditioning has historically been a popular method of psychotherapy (i.e., systematic de-
sensitization; e.g., Wolpe, 1958), especially for the treatment of phobias and anxiety. In
counterconditioning-based therapies, a stimulus (e.g., a spider) previously associated
with an unpleasant emotional reaction (e.g., fear) is now associated with a pleasant emo-
tional reaction (e.g., relaxation). Responding to the stimulus is usually a blend of the two
associations, thereby reducing or eliminating the fear reaction. We must note, however,
that current approaches to phobia and anxiety treatment prefer an approach different
from counterconditioning. Most contemporary clinicians use exposure therapies, which
are based on experimental extinction. In these therapies, the stimulus that elicits the un-
pleasant emotional reaction (in our previous example, the spider) is now presented and
the client is allowed to experience the lack of an outcome (e.g., is not bitten by the spi-
der), with the usual consequence that the unpleasant emotional reaction undergoes extinc-
tion and its intensity gradually decreases. The difference between the two approaches is
that, while counterconditioning uses training with a different outcome as the therapy pro-
cedure (i.e., the therapy sessions), extinction uses training with no outcome as the therapy
procedure (i.e., it degrades the contingency between the stimulus and the emotional reac-
tion).
Types of Stimulus Interference in Basic Learning
Figure 1. A 2 x 2 matrix depicting the different types of stimulus interference that can disrupt acquired responding. In each cell, representative examples of inter-
ference procedurally appropriate for that cell are listed. ‘X‘ represents the target conditioned stimulus, ‘A‘ represents the interfering cue, ‘O1‘ and ‘O2’ represent
distinctly different outcomes, which might be unconditioned stimuli (USs) or innocuou s stimuli that are later paired with USs. Th e bold font of th e overshadowing
cue in Cell 1 (denoted as ‘A’) reflects the finding that overshadowing of a target cue is greatest when the overshadowing cue is con siderably m ore salient than the
target cue. RI = retroactive interference, PI = proactive interference. Contemporary models of acquired behavior have focused almost exclusively on Cell 1 phe-
nomena to the exclusion of phenomena in Cells 2, 3, and 4. See text for elaboration
Competing Cues Competing Outcomes
Trained Together
Trained Apart
Overshadowing (Pavlov, 1927):
AXO1, Test XO1
Blocking (Kamin, 1968):
AO1, then AXO1, Test XO1
Relative validity (Wagner et al., 1968):
AXO1, BXnoO1, Test XO1
Overexpectation (Rescorla, 1970):
XO1, AO1, then XAO1, Test XO1
Rescorla (1980)
Esmoris-Arranz et al. (1997)
Miller & Matute (1998)
XO1, then XO1+O2, Test XO2
Matute & Pineño (1998)
Escobar, Matute, & Miller (2001)
Escobar, Arcediano, & Miller (2001)
XO1, AO1, Test XO1 (RI)
AO1, XO1, Test XO1 (PI)
XO1, XO2, Test XO1 (RI)
XO2, XO1, Test XO1 (PI)
Counterconditioning (Pavlov, 1927):
XUS1, XUS2, Test XUS1
Cell 1: Cell 2:
Cell 3: Cell 4:
O1
Cue A
Cue X
O1
O1
Cue X
Cue X
O2
Cue A
Cue X
O1
Cue X
O1
O2
-293-
Recency and Primacy Effects in Interference Situations
Cell 3- and Cell 4-type interference effects are also sensitive to time. Usually, if
testing occurs immediately after training, subjects exhibit conditioned responding consis-
tent with the more recent training (i.e., recency). However, this recency bias wanes as the
retention interval increases, often unmasking a primacy effect (i.e., with increasing reten-
tion interval, retroactive interference decreases and proactive interference increases; e.g.,
Postman et al., 1968).
Interference vs. Degraded Contingency
Note that situations in which two cues are trained with one outcome (Cell 3) and
situations in which one cue is trained with two outcomes (Cell 4) could be explained in
terms of degradation of the cue-outcome contingency: The second stage of treatment
represents a situation in which one of the elements used during the first stage of treatment
is presented without the other. However, recent reports suggest that the effect of interfer-
ence training is significantly greater than the effect of contingency degradation alone
(Escobar, Arcediano, & Miller, 2001; Escobar, Matute, & Miller, 2001). Note that our
working definition of interference differs from other definitions of interference (e.g.,
Bouton, 1993) in which interference refers to both what we here call degraded contin-
gency effects and interference effects. Degraded contingency effects arise from present-
ing the cue without the outcome or presenting the outcome without the cue. Although no-
outcome representations and no-cue representations may well serve as an effective asso-
ciate, thereby transforming degraded contingency situations into interference situations,
this is a theoretical issue that we wish to circumvent in this discussion.
Effect of Retrieval Cues
Importantly, interference has been considered one of the major sources of forget-
ting (e.g., McGeoch, 1932). Forgetting due to interference is usually assumed to reflect a
temporary (rather than permanent) inaccessibility to the information stored in memory.
This assumption is supported by the observation that several manipulations performed at
the time of testing can attenuate or generate interference (e.g., Spear, 1973). For exam-
ple, Escobar, Matute, and Miller (2001) trained rat subjects in an interference preparation
such that retrieval of a target cue was impaired. They observed that presenting a cue that
was present during training of the target cue attenuated interference and similar effects
were observed by placing the animal back in the environment of target training. Con-
versely, presenting a cue that was present during training with the interfering cue pro-
duced interference in a situation in which no interference was otherwise observed.
Interference vs. Facilitation
We have described different types of interference as the ubiquitous effects of the
four training situations described in Figure 1. However, sometimes training consistent
with the four cells of Figure 1 results in facilitation (i.e., increased conditioned respond-
ing) rather than interference. Cell 1-type treatments sometimes result in the so-called
“potentiation” effect. For example, subjects presented with the compound of a taste and
-294-
an odor followed by injection of a drug that induces internal malaise fail to exhibit com-
petition between the two cues for prediction of the malaise. Rather, the aversion to the
odor (or the taste) is increased relative to subjects that receive pairings of the odor (or
taste) alone with the drug (Clarke, Westbrook, & Irwin, 1979; Rusiniak, Hankins, Gar-
cia, & Brett, 1979; see Batsell & Batson, 1999, for a similar effect using a blocking pro-
cedure). Cell 3- and Cell 4-type treatments sometimes result in acquired equivalence (or
acquired similarity), in which subjects respond to two stimuli equivalently because they
were previously paired with the same outcome. For example, Honey and Hall (1989)
trained two cues, A and B, as signals for a common outcome (a food pellet). Then, A was
paired with shock. When B was presented at test, subjects exhibited substantial generali-
zation of the fear conditioned response. That is, the common training history shared by A
and B rendered them functionally equivalent.
Effects of Similarity on Interference
Observing stimulus interference requires that there be a certain level of similarity
between the two associations (e.g., sharing a common element). However, if the two as-
sociations are identical, summation of learning rather than interference will be observed
(e.g., Young, 1955). Similarly, if the two associations are completely dissimilar, exceed-
ingly weak interference is usually observed (e.g., Escobar & Miller, 2003; Newton &
Wickens, 1956).
Similarity of Training to Test Cue
Variation in the external world and inside the subject necessarily results in
changes in the perceived characteristics of a stimulus over repeated presentations. Thus,
it is necessary that the subject perceive slight variations of the same stimulus as instances
of a previously experienced stimulus category. That is, the process of stimulus generali-
zation is critical to the observation of any learning at all. In general, the maximum
amount of conditioned responding is observed to the cue with which subjects were
trained, and responding to other cues decreases as the similarity between them and the
target cue decreases. This similarity-based decrease in conditioned responding is known
as generalization decrement (e.g., Guttman & Kalish, 1956).
Motivation
Motivation is a complex topic with a huge literature that we could not begin to
detail here. In the framework of learning, we talk about stimuli with motivational value to
refer to stimuli that elicit a response because they are of either inherent biological signifi-
cance (primary reinforcers; e.g., food, sex, pain) or acquired biological significance (sec-
ondary [learned] reinforcers; e.g., money, praise). A great body of research now suggests
that the process of learning does not require the presence of a biologically significant
stimulus; that is, learning can occur between neutral cues. Learning between neutral
stimuli can be observed in both classical conditioning (sensory preconditioning; Brodgen,
1959) and instrumental conditioning (latent learning; Tolman & Honzik, 1930). How-
ever, regardless of what relationships have been encoded by a subject, learning is not
observed unless the outcome, or an associate of the outcome, has the potential to elicit a
-295-
response in its own right. That is, motivational value of one of the associates is necessary
for the expression of learning. For example, think of an instance of latent learning in
which rats were allowed to walk through a maze without food reinforcement in the goal
box. The rats show little improvement in reaching the goal box over trials, so it is not
obvious that they learned about the maze. However, when food reinforcement is intro-
duced, the rats that had previously been exposed to the maze perform better than rats that
had not been allowed to interact with the maze before (Tolman & Honzik, 1930). That is,
when exposed to the maze in the absence of reward, the rats acquired information about
the maze, but they did not express that learning until a biologically significant stimulus
entered the equation.
Extension of These Laws to Other Forms of Learning
The present review has focused on selected empirical evidence from the classical
conditioning literature. However, most of the laws described here apply to other forms of
learning, such as instrumental conditioning and causality learning (e.g., Allan, 1993;
Miller & Balaz, 1981). Description of all the literature pertaining to these forms of learn-
ing and the laws described here would require a book. Thus, we will limit our discussion
to highlighting some empirical findings to emphasize the generality of the laws we have
described.
Salience of the cue and outcome are certainly known to influence instrumental
performance. Of special interest are the effects of outcome salience, which can be quanti-
fied in terms of quantity and quality of the reinforcer. In general, animals respond more
for larger, more palatable reinforcers (e.g., Hutt, 1954), and changes in reinforcer quan-
tity result in marked changes in behavior (so-called contrast effects, e.g., Crespi, 1942).
Mediation is also an issue in instrumental conditioning, as exemplified by secondary or
conditioned reinforcement, in which a stimulus previously paired with a primary (i.e.,
biologically significant) reinforcer comes to act as a reinforcer in its own right (for dis-
cussion, see Williams, 1994). Conditioned reinforcement has also been observed in hu-
man causality judgments. For example, Reed (1999) reported that a signal presented be-
tween a response (pressing a key) and an outcome (a triangle lighting up in a computer
screen) came to act as a conditioned reinforcer. Negative mediation has also been repeat-
edly reported in the human causality learning literature. For example, Chapman (1991)
observed conditioned inhibition of causality judgments in a task in which subjects were
asked to rate the likelihood that a given symptom was indicative of developing a disease.
It has been long known that delays as short as 0.5 s between the response and the
outcome are known to adversely affect instrumental responding (e.g., Grice, 1948), and
this deleterious effect is directly related to the length of the delay (e.g., Dickinson, Watt,
& Griffiths, 1992). That is, response-outcome contiguity is a relevant factor in instru-
mental conditioning. Furthermore, consistent with the observation of inhibition of delay
in Pavlovian conditioning, animals delay the onset of their instrumental responding to
approximate the time of reinforcement availability (e.g., Skinner, 1938). Such exquisite
timing of responding has also been observed in human conditioning (e.g., Arcediano,
Escobar, & Miller, 2003).
The role of predispositions in instrumental responding has been long known. For
example, Foree and LoLordo (1973, 1975) found that pigeons could more readily be
trained to peck for food in the presence of a visual cue and treadle press (i.e., part of a
-296-
running response) to avoid shock in the presence of an auditory cue than with other
stimulus-outcome combinations. This work parallels Garcia and Koelling’s (1966)
“bright-noisy water” study in Pavlovian conditioning. Thus, predispositions are observed
in instrumental as well as Pavlovian situations, which was anticipated by Thorndike’s
(1911) concept of “readiness.” Predispositions are also observed in human causality
learning. For example, Bullock, Gelman, & Baillargeon (1982) have argued that humans
are predisposed to acquire causal relations between events in their environment even at a
young infant age. More recently, Waldmann (e.g., 2000) has argued that humans are pre-
disposed to learn causal relations in a cause-to-effect direction, and that this direction
cannot be reversed (but there are multiple detractors of this view; see e.g., López et al.,
1998; Matute, Arcediano, & Miller, 1996).
The effects of contingency on instrumental reinforcement are obvious: If there is
no contingency between the response and the reinforcer, instrumental responding is dis-
rupted (i.e., extinction). Partial reinforcement produces one of the best-known contin-
gency effects. Partially-reinforced behavior is more resistant to extinction than continu-
ously-reinforced behavior (the partial reinforcement extinction effect). But less attention
has been paid to the unsurprising but important observation that partial reinforcement
usually attenuates instrumental responding (e.g., Jenkins & Stanley, 1950). Interference
effects are also obvious in instrumental conditioning, although not all cells of Figure 1
have been explored. For example, Cell 1-type interference has been observed in the form
of blocking (e.g., Hammerl, 1993; Roberts & Pearce, 1999). More extensive research has
been performed in human causality judgments. Cell 1-type interference has been ob-
served in blocking (e.g., Arcediano, Matute, & Miller, 1997; Arcediano, Escobar, &
Matute, 2001) and overshadowing (e.g., Price & Yates, 1993). Cell 2-type interference
has been observed in selected occasions (e.g., Matute, Arcediano, & Miller, 1996;
Shanks & López, 1996; but see e.g., Waldmann, 2000; Waldmann & Holyoak, 1992).
Cell 3- and Cell 4-type interference has been extensively reported in the human verbal
learning tradition (for reviews, see e.g., Britt, 1935; Slamecka & Ceraso, 1960; Swenson,
1941), and more recently in the human causality learning literature (e.g., Escobar,
Pineño, & Matute, 2002). As mentioned in the section on Effects of Similarity on Inter-
ference, decreased similarity between two outcomes can enhance stimulus control in se-
lect situations. For example, setting a differential outcomes schedule, in which different
reinforcers follow each of two (or more) potential choice behaviors, results in faster dis-
crimination than if the same reinforcer was given for both behaviors (see Overmier, Sav-
age, & Sweeney, 1999, for a review).
Summary
Table 3 summarizes the conditions that favor the acquisition of stimulus control
of behavior. These conditions are loosely ranked according to our impression of their
relative importance (except Generalization and Motivation, which speak to test rather
than acquisition conditions). Whether the conditions in Table 3 are viewed as influencing
the expression of previously acquired relationships (Denniston, Savastano, & Miller,
2001; Miller & Matzel, 1988) or the fostering of new learning about absent stimuli
(Dickinson & Burke, 1996; Van Hamme & Wasserman, 1994) is a source of current con-
troversy in the literature.
-297-
Table 3.
Summary of the Conditions that Favor Learning.
Condition Effect on learning
Stimulus salience High salience of the cue results in faster acquisition of responding; high
salience of the outcome results in a higher asymptote of responding.
Similarity and con-
tiguity Similarity between the cue and outcome facilitates learning. Spatial and
temporal contiguity (direct or mediated) of the cue and outcome also
facilitates learning. Fast development of responding results from massed
trials, but long-term retention results from distributed (i.e., spaced) trials.
Predispositions Animals are predisposed to associate certain cues to certain outcomes.
Selection of a cue and outcome that have (or had) biologically important
relationships in the subject’s current (or ancestral) ecological niche fa-
cilitates learning.
Contingency Large numbers of trials that confirm the cue-outcome relationship (i.e.,
trials in which the two events occur and trials in which the two events do
not occur) relative to the number of trials that disconfirm this relation-
ship (i.e., trials in which the cue or outcome occur alone) facilitate learn-
ing.
Interference Learning of a target cue-outcome relationship benefits from situations in
which there are few alternative cues paired with the outcome and few
alternative outcomes paired with the target cue.
Generalization Responding benefits from high similarity between the training and test-
ing cues.
Motivation Even though motivation is not necessary for learning to occur, it is nec-
essary for the expression of learned associations. The biological signifi-
cance of an outcome depends on both the physical properties of the out-
come and the state of the subject.
We conclude that theory has shifted the focus from some of the empirical condi-
tions that favor (or impair) the development of learning. This is not to deny that theories
of learning at the psychological level have important beneficial functions. They can: (a)
predict behavior, (b) be used to control behavior, (c) organize empirical phenomena, and
(d) have heuristic value in inspiring illuminating experiments that reveal previously un-
recognized empirical relationships (Platt, 1964; Popper, 1959). However, in contrast to
the prevailing arguments concerning appropriateness of various models, there is a rather
clear set of principles (laws) that describe the conditions favoring acquired behavior,
even if we do not have a model at this time which accounts for all of these principles.
Importantly, optimal conditions for learning clearly vary as a function of the goals of the
task at hand (e.g., rapidity of acquisition vs. resistance interference). Anyone interested in
learning theory must keep these principles in mind; and researchers who are not inter-
ested in psychological theories of learning, but rather are concerned with the physiology
-298-
of learning, or the application of learning to clinical or educational problems, would be
well advised to attend to these principles, rather than exclusively to the various contem-
porary models that have been proposed to account for these basic phenomena.
References
Allan, L. G. (1993). Human contingency judgments: Rule based or associative? Psychological
Bulletin, 114, 435-448.
Amun dson, J . C., E scobar, M., & Miller , R. R. (2003). Proactive interference between cues trained
with a common outcome in first-order Pavlovian conditioning. Journal of Experimental Psychology: Animal
Behavior Processes, 29, 311-322.
Arcediano, F., Escobar, M., & Matute, H. (2001). Reversal from blocking in humans as a result of
posttraining extinction of the blocking stimulus. Animal Learning and Behavior, 29, 354-366.
Arcedia no, F., E scobar, M., & Miller, R. R. (2003). Temporal integration and temporal backward
associations in human and nonhuman subjects. Learning and Behavior, 31, 242-256.
Arcedia no, F., Matu te, H., & Miller , R. R. (1997). Blocking of Pavlovian conditioning in humans.
Learning and Motivation, 28, 188-199.
Batsell, W. R., Jr., & Batson, J. D. (1999). Augmentation of taste conditioning by a preconditioned
odor. Journal of Experimental Psychology: Animal Behavior Processes, 25, 374-388.
Berkeley, G. (1710/1946). A treatise concerning the principles of human knowledge. La Salle, IL:
Open Court Publication Co. (Reprinted from Dublin, Ireland: Jeremy Pepyat).
Bjork, R. A. (1988). Retrieval practice and the maintenance of knowledge. In M. M. Gruneberg, P.
E. Morris, & R. N. Sykes (Eds.), Practical aspects of memory: Current research and issues (Vol. 1, pp.
296-401). Chichester, UK: John Wiley & Sons.
Bouton, M. E. (1993). Context, time, and memory retrieval in interference paradigms of Pavlovian
learning. Psychological Bulletin, 114, 80-99.
Bouton, M. E., & King, D. A. (1983). Contextual control of the extinction of conditioned fear:
Tests for the associative value of the context. Journal of Experimental Psychology: Animal Behavior Proc-
esses, 9, 248-265.
Britt, S. H. (1935). Retroactive inhibition: A review of the literature. Psychological Bulletin, 32,
381-440. Brogden, W. J. (1939). Sensory pre-conditioning. Journal of Experimental Psychology, 25, 323-
332. Bullock, M., Gelman, R., & Baillargeon, R. ( 1982). The development of causal reasoning. In W. J.
Friedman (Ed.), The developmental psychology of time (pp. 209-254). New York: Academic Press.
Burns, M., & Domjan, M. (1996). Sign tracking versus goal tracking in the sexual conditioning of
male Japanese quail (Coturnix japonica). Journal of Experimental Psychology: Animal Behavior Processes,
22, 297-306.
Cameron, C. L., Cella, D., Herndon, J. E. II, Kornblith, A. B., Zuckerman, E., Henderson, E.,
Weiss, R. B., Cooper, M. R., Silver, R. T., Leone, L., Canellos, G. P., Peterson, B. A., & Holland, J. C.
(2001). Persistent symptoms among survivors of Hodgkin’s disease: An explanatory model based on classical
conditioning. Health Psychology, 20, 71-75.
Chang, R. C., Stout, S., & Miller, R. R. ( 2004). Comparing excitatory backward and forward con-
ditioning. Quarterly Journal of Experimental Psychology, 57B, 1-23.
Channell, S., & Hall, G. (1983). Contextual effects in latent inhibition with an appetitive
conditioning procedure. Animal Learning and Behavior, 11, 67-74.
Chapman, G. B. (1991). Trial order affects cue interaction in contingency judgment. Journal of
Experimental Psychology: Learning, Memory and Cognition, 17, 837-854.
Clarke, J. C., Westbrook, R. F., & Irwin, J. (1979). Potentiation instead of overshadowing in the
pigeon. Behavioral and Neural Biology, 25, 18-29.
Cook, M., & Mineka, S. (1990). Selective associations in the observational conditioning of fear in
monkeys. Journal of Experimental Psychology: Animal Behavior Processes, 16, 372-389.
Cook, T. W. (1934). Massed and distributed practice in puzzle solving. Psych ologica l Review, 41,
330-355. Crespi, L. P. (1942). Quantitative variation in incentive and performance in the white rat.
-299-
American Journal of Psychology, 55, 467-517.
Dalrymple, A. J., & Galef, B. G., Jr. (1981). Visual discrimination pretraining facilitates subse-
quent visual cue/toxicosis conditioning in rats. Bulletin of the Psychonomic Society, 18, 267-270.
De la Casa, L. G., & Lubow, R. E. (2002). An empirical analysis of the super-latent inhibition
effect. Animal Learning and Behavior, 30, 112-120.
Denniston, J. C., Savastano, H. I., & Miller, R. R. (2001). The extended comparator hypothesis:
Learning by contiguity, responding by relative strength. In R. R. Mowrer & S. B. Klein (Eds.), Handbook of
contemporary learning theories (pp. 65-117). Hillsdale, NJ: Erlbaum.
Dickinson, A., & Burke, J. (1996). Within-compound associations mediate the retrospective
revaluation of causality judgments. Quarterly Journal of Experimental Psychology, 49B, 60-80.
Dickinson, A., Watt, A., & Griffiths, W. J. H. (1992). Free-operant acquisition with delayed rein-
forcement. Quarterly Journal of Experimental Psychology, 45B, 241-258.
Ebbinghaus, H. (1885/1913). Memory: A contribution to experimental psychology. (H. A. Ruger
& C. E. Bussenius, Trans.). New York: Columbia University Teacher’s College.
Escobar, M., & Miller, R. R. (2003). Timing in retroactive interference. Learning and Behavior,
31, 257-272.
Escobar, M., Arcediano, F., & Miller, R. R. (2001). Conditions favoring retroactive interference
between antecedent events and between subsequent events. Psychonomic Bulletin and Review, 8, 691-697.
Escobar, M., Matut e, H., & Miller, R. R. (2001). Cues trained apart compete for behavioral control
in rats: Convergence with the associative interference literature. Journal of Experimental Psychology:
General, 130, 97-115.
Escobar, M., Pineño, O., & Matute, H. (2002). A comparison between elemental and compound
training of cues in retrospective revaluation. Animal Learning and Behavior, 30, 228-238.
Esm oris-Arranz, F. J., Miller, R. R., & Matut e, H. (1997). Blocking of subsequent and antecedent
events. Journal of Experimental Psychology: Animal Behavior Processes, 23, 145-156.
Esmoris-Arranz, F. J., Pardo-Vázquez, J. L., & Vázquez-García, G. A. (2003). Differential effects
of forward or simultaneous conditioned stimulus-unconditioned stimulus intervals on the defensive behavior
system of the Norway rat (Rattus norvegicus). Journal of Experimental Psychology: Animal Behavior
Processes, 29, 334-340.
Estes, W. K. (1950). Toward a statistical theory of learning. Psychological Review, 57, 94-170.
Foree, D. D., & LoLordo, V. M. (1973). Relation of cue to consequence in avoidance learning.
Psychonomic Science, 4, 123-124.
Foree, D. D., & LoLordo, V. M. (1975). Stimulus-reinforcer interactions in the pigeon: The role of
electric shock and the avoidance contingency. Journal of Experimental Psychology: Animal Behavior Proc-
esses, 1, 39-46.
Garcia, J., & Koelling, R. A. (1966). Relation of cue to consequence in avoidance learning. Psy-
chonomic Science, 4, 123-124.
Garcia, J., E rvin, F. R ., & Koelling, R. A. (1966). Learning with prolonged delay of reinforcement.
Psychonomic Science, 5, 121-122.
Gibbon, J., & Balsam, P. (1981). Spreading association in time. In C. M. Locurto, H. S. Terrace, &
J. Gibbon (Eds.), Autoshaping and conditioning theory (pp. 219-253). New York: Academic Press.
Gibbon, J., Baldock, M. D., Locurto, C., Gold, L., & Terrace, H. S. (1977). Trial and intertrial du-
rations in autoshaping. Journal of Experimental Psychology: Animal Behavior Processes, 3, 264-284.
Grice, G. R. (1948). The relation of secondary reinforcement to delayed reward in visual discrimi-
nation learning. Journal of Experimental Psychology, 38, 1-16.
Guthrie, E. R. (1935). The psychology of learning. New York: Harper.
Guttman, N., & Kalish, H. I. (1956). Discriminability and stimulus gen eralization. Journal of Ex-
perimental Psychology, 51, 79-88.
Gwinn, G. T. (1949). The effects of punishment on acts motivated by fear. Journal of Experimen-
tal Psychology, 39, 260-269.
Hammerl, M. (1993). Blocking observed in human instrumental conditioning. Learning and Mo-
tivation, 24, 73-87.
Heth, C. D. (1976). Simultaneous and backward fear conditioning as a function of number of CS-
UCS pairings. Journal of Experimental Psychology: Animal Behavior Processes, 2, 117-129.
Holland, P.C. (1992). Occasion setting in Pavlovian conditioning. In D.L. Medin (Ed.), The Psy-
chology of learning and motivation (Vol. 28, pp. 69-125). San Diego, CA: Academic Press.
Hollis, K. L. (1997). Contemporary research on Pavlovian conditioning. American Psychologist,
-300-
52, 956-965.
Hollis, K. L., Cadieu x, E. L., & Colbert, M. M. (1989). The biological function of Pavlovian condi-
tioning: A mechanisms for mating success in the blue gourami (Trichogaster trichopterus). Journal of Com-
parative Psychology, 103, 115-121.
Honey, R. C., & Hall, G. (1989). Acquired equivalence and distinctiveness of cues. Journal of Ex-
perimental Psychology: Animal Behavior Processes, 15, 338-346.
Hovland, C. I. (1940). Experimental studies in rote-learning theory. VI. Comparison of retention
following learning to same criterion by massed and distributed practice. Journal of Experimental Psychol-
ogy, 26, 568-587.
Hull, C. L. (1952). A behavior system. New Haven: Yale University Press.
Hutt, P. J. (1954). Rate of bar pressing as a function of quality and quantity of food reward. Jour-
nal of Comparative and Physiological Psychology, 47, 235-239.
Jenkins, W. O., & Stanley, J. C., Jr. (1950). Partial reinforcement: A review and a critique. Psy-
chological Bulletin, 47, 193-204.
Kamin, L. J. (1965). Temporal and intensity characteristics of the conditioned stimulus. In W. F.
Prokasy (Ed.), Classical conditioning (pp. 118-147). New York: Appleton-Century-Crofts.
Kamin, L. J. (1968). "Attention-like" processes in classical conditioning. In M. R. Jones (Ed.)
Miami Symposium on the Prediction of Behavior: Aversive stimulation (p p. 9 -3 1). Miami, FL : Univer sity of
Miami Press.
Kao, S. F., & Wasserman, E. A. (1993). Assessment of an information integration account of con-
tingency judgment with examination of subjective cell importance and method of information presentation.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1363-1386.
Kimble, G. A. (1947). Conditioning as a function of the time between conditioned and uncondi-
tioned stimuli. Journal of Experimental Psychology, 37, 1-15.
Kimble, G. A. (1955). Shock intensity and avoidance learning. Journal of Comparative and
Physiological Psychology, 48, 281-284.
Kraemer, P. J., Randall, C. K., & Carbary, T. J. (1991). Release from latent inhibition with delayed
testing. Animal Learning and Behavior, 19, 139-145.
Krane, R. V., & Wagner, A. R. (1975). Taste aversion learning with a delayed shock US: Implica-
tions for the "generality of the laws of learning." Journal of Comparative and Physiological Psychology, 88,
882-889. Landauer, T. K., & Bjork, R. A. (1978). Optimal rehearsal patterns and name learning. In M. M.
Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical aspects of memory (pp. 625-632). New York:
Academic Press.
LoLordo, V. M., & Fairless, J. L. (1981). Pavlovian conditioned inhibition: The literature since
1969. In R. R. Miller and N. E. Spear (Eds.), Information processing in animals: Conditioned inhibition
(pp. 1-49). Hillsdale, NJ: Erlbaum.
López, F. J., Shanks, D. R., Almaraz, J., & Fernández, P. (1998). Effects of trial order on contin-
gency judgments: A compariso n of asso ciative an d probabilistic contrast accounts. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 24, 672-694.
Lubow, R.E., & Moore, A.U. (1959). Latent inhibition: The effect of nonreinforced preexposure to
the conditioned stimulus. Journal of Comparative and Physiological Psychology, 52, 415-419.
Matute, H., & Pineño, O. (1998). Stimulus competition in the absence of compound conditioning.
Animal Learning and Behavior, 26, 3-14.
Matute, H., Arcediano, F., & Miller, R. R. (1996). Test question modulates cue competition
between causes and between effects. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 22, 182-196.
Matzel, L. D., Held, F., & Miller, R.R. (1988). Information and expression of simultaneous and
backward associations: Implications for contiguity theory. Learning and Motivation, 19, 317-344.
McAllister, W. R. (1953). Eyelid conditioning as a function of the CS-UCS interval. Journal of
Experimental Psychology, 45, 417-422.
McGeoch, J. A. (1932). Forgetting and the law of disuse. Psychological Review, 39, 352-370.
Miller, R. R., & Balaz, M . A. (1981). Differences in adaptiveness between classically conditioned
responses an d instrumentally acq uired respon ses. In N. E . Spear & R . R. Miller (E ds.), Info rmation process-
ing in animals: Memory mechanisms (pp. 49-80). Hillsdale, NJ: Erlbaum.
Miller, R. R., & Escobar, M. (2002). Learning: Laws and models of basic conditioning. In C. R.
Gallistel (Ed.), Stevens’ handbook of experimental psychology. Vol.3: Learning, motivation, and emotion.
-301-
New York: John Wiley & Sons.
Miller, R. R., & Matute, H. (1996). Biological significance in forward and backward blocking:
Resolution of a discrepancy between animal conditioning and human causal judgment. Journal of Experi-
mental Psychology: General, 125, 370-386.
Miller, R. R., & Matute, H. (1998). Competition between outcomes. Psychological Science, 9,
146-149. Miller, R. R., & Matzel, L. D. (1988). The comparator hypothesis: A response rule for the
expression of associations. In G. H. Bower (Ed.), The psychology of learning and motivation, Vol. 22, (pp.
51-92). San Diego, CA: Academic Press.
Miller, R. R., & Oberling, P. (1998). Analogies between occasion setting and Pavlovian
conditioning. In N. A. Schmajuk and P. C. Holland (Eds.), Occasion setting: Associative learning and
cognition in animals (pp. 3-35). Washington, DC: American Psychological Association.
Miller, R. R., Hallam, S. C., Hong, J. Y., & Dufore, D . S. (1991). Associative structure of differen-
tial inhibition: Implications for models of conditioned inhibition. Journal of Experimental Psychology: Ani-
mal Behavior Processes, 17, 141-150.
Miller, R. R., Kasprow, W. J ., & Schachtman, T . R. (1985). Retrieval variability: Sources and con-
sequences. American Journal of Psychology, 99, 145-218.
Mineka, S., Davidson, M., Cook, M., & Keir, R. (1984). Observational conditioning of snake fear
in rhesus monkeys. Journal of Abnormal Psychology, 93, 355-372.
Newton, J. M., & Wickens, D. D. (1956). Retroactive inhibition as a function of the temporal posi-
tion of the interpolated learning. Journal of Experimental Psychology, 51, 149-154.
Overmier, J. B., Savage, L. M., & Sweeney, W. A. (1999). Behavioral and pharmacological analy-
ses of memory: New behavioral options for remediation. In M. Haugh & R. E. Whalen (Eds.), Animal mod-
els of human emotio n and cognition (pp. 231-245). Washington, DC: American Psychological Association.
Pavlov (1927). Conditioned reflexes. Oxford, UK: Oxford University Press.
Pearce, J. M., Redhead, E. S., & Aydin, A. (1997). Partial reinforcement in appetitive Pavlovian
conditioning in rats. Quarterly Journal of Experimental Psychology, 50B, 273-294.
Platt, J. R. (1964). Strong inference. Science, 146, 347-353.
Popper, D. R. (1959). The logic of scientific discovery. New York: Basic Books.
Postman, L. (1962). Transfer of training as a function of experimental paradigm and degree of
first-list learning. Journal of Verbal Learning and Verbal Behavior, 1, 109-118.
Postman, L., Stark, K., & Fraser, J. (1968). Temporal changes in interference. Journal of Verbal
Learning and Verbal Behavior, 7, 672-694.
Price, P. C., & Yates, J. F. (1993). Judgmental overshadowing: Further evidence of cue competi-
tion in contingency judgments. Memory and Cognition, 21, 561-572.
Randich, A., & LoLordo, V. M. (1979). Preconditioning exposure to the unconditioned stimulus
affects the acquisition of a conditioned emotional response. Learning and Motivation, 10, 245-277.
Rashotte, M. E. (1981). Second-order autoshaping: Contributions to the research and theory of
Pavlovian reinforcement by conditioned stimuli. In C. M. Locurto, H. S. Terrace, & J. Gibbon (Eds.), Auto-
shaping and conditioning theory (pp. 139-180). New York: Academic Press.
Reed, P. (1999). Role of a stimulus filling an action-ou tcome delay in human judgments of causal
effectiveness. Journal of Experimental Psychology: Animal Behavior Processes, 25, 92-102.
Rescorla, R. A. (1967). Inhibition of delay in Pavlovian fear conditioning. Journal of Comparative
and Physiological Psychology, 64, 114-120.
Rescorla, R. A. (1968). Probability of shock in the presence and absen ce of CS in fear condition-
ing. Journal of Comparative and Physiological Psychology, 66, 1-5.
Rescorla, R. A. (1969). Pavlovian conditioned inhibition. Psychological Bulletin, 72, 77-94.
Rescorla, R. A. (1970). Reduction in the effectiveness of reinforcement after prior excitatory con-
ditioning. Learning and Motivation, 1, 372-381.
Rescorla, R. A. (1980). Pavlovian second-order conditioning. Hillsdale, NJ: Erlbaum.
Rescorla, R. A. (1999). Within-subject partial reinforcement extinction effect in autoshaping.
Quarterly Journal of Experimental Psychology, 52B, 75-87.
Rescorla, R. A., & Cunningham, C. L. (1979). Spatial contiguity facilitates P avlovian secon d-order
conditioning. Journal of Experimental Psychology: Animal Behavior Processes, 5, 152-161.
Rescorla, R. A., & Furrow, D. R. (1977). Stimulus similarity as a determinant of Pavlovian condi-
tioning. Journal of Experimental Psychology: Animal Behavior Processes, 3, 203-215.
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the
-302-
effectiveness of reinforcement and non-reinforcement. In A. H. Black and W. F. Prokasy (Eds.), Classical
conditioning II: Current theory and research (pp. 64-99). New York: Appleton-Century Crofts.
Roberts, A. D. L, & Pearce, J. M. (1999). Blocking in the Morris swimming pool. Journal of Ex-
perimental Psychology: Animal Behavior Processes, 25, 225-235.
Rom aniuk, C. B ., & Williams, D. A. (2000). Conditioning across the duration of a backward con-
ditioned stimulus. Journal of Experimental Psychology: Animal Behavior Processes, 26, 454-461.
Rusiniak, K. W., Hankins, W. G., Garcia, J., & Brett, L. P. (1979). Flavor-illness aversions: Poten -
tiation of odor by taste in rats. Behavioral and Neural Biology, 25, 1-17.
Savastan o, H. I., & Miller, R. R. (1998). Time as content in Pavlovian conditioning. Behavioural
Processes, 44, 147-162.
Savastano, H. I., Cole, R. P., Barnet, R. C., & Miller, R. R. (1999). Reconsidering conditioned
inhibition. Learning and Motivation, 30, 101-127.
Shanks, D. R., & Lopez, F. J. (1996). Causal order does not affect cue selection in human
associative learning. Memory and Cognition, 24, 511-522.
Sherrington, C. S. (1947). The integrative action of the central nervous system. Cambridge:
Cambridge University Press.
Skinner, B. F. (1938). The behavior of organisms: An experimental analysis. Englewood Cliffs,
NJ: Prentice-Hall.
Skinner, B. F. (1950). Are learning theories necessary? Psychological Review, 57, 193-216.
Slamecka, N. J., & Ceraso, J. (1960). Retroactive and proactive inhibition of verbal learning.
Psychological Bulletin, 57, 449-475.
Spear, N. E. (1973). Retrieval of memory in animals. Psychological Review, 80, 163-194.
Stockhorst, U., Klosterhalfen, S., & Steingrueber, H-. J. (1998). Conditioned nausea and further
side-effects in cancer chemotherapy: A review. Journal of Psychophysiology, 12, 14-33.
Swenson, E. J. (1941). Retroactive inhibition: A review of the literature. University of Minnesota
Studies in Education (Vol. 1). Minneapolis: University of Minnesota Press.
Testa, T. J. (1975). Effects of similarity of location and temporal intensity pattern of conditioned
and unconditioned stimuli on the acquisition of conditioned suppression in rats. Journal of Experimental
Psychology: Animal Behavior Processes, 1, 114-121.
Thorndike, E. L. (1898). Animal intelligence : An ex perimen tal st udy of the associative processes in
animals. Psychological Monographs, 2.
Thorndike, E. L. (1911). Animal intelligence: Experimental studies. New York: Macmillan.
Thorndike, E. L. (1932). Fundamentals of learning. New York: Columbia University.
Timberlake, W., & Lucas, G. A. (1991). Periodic water, interwater interval, and adjunctive behav-
ior in a 24-hour multiresponse environment. Animal Learning and Behavior, 19, 369-380.
Tolman, E. C., & Honzik, C. H. (1930). Introduction and removal of reward, and maze perform-
ance in rats. University of California Publications in Psychology, 4, 257-275.
Underwood, B. J. (1948). 'Spontaneous recovery' of verbal associations. Journal of Experimental
Psychology, 38, 429-439.
Van Hamme, L. J., & Wasserman, E. A. (1994). Cue competition in causality judgments: The role
of nonpresentation of compound stimulus elements. Learning and Motivation, 25, 127-151.
Wagner, A. R., Logan, F. A., Haberlandt, K., & Price, T. (1968). Stimulus selection in animal dis-
crimination learning. Journal of Comparative and Physiological Psychology, 76, 171-180.
Waldmann, M. R. (2000). Competition among causes but not effects in predictive and diagnostic
learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 53-76.
Waldmann, M. R., & Holyoak, K. J. (1992). Predictive and diagnostic learning with causal models:
Asymmetries in cue competition. Journal of Experimental Psychology: General, 121, 222-236.
Wasserman, E. A., Franklin, S. R., & Hearst, E. (1974). Pavlovian appetitive contingencies and
approach versus withdrawal to conditioned stimuli in pigeons. Journal of Comparative and Physiological
Psychology, 86, 616-627.
Williams, B. A. (1994). Conditioned reinforcement: Neglected or outmoded explanatory construct?
Psychonomic Bulletin and Review, 1, 457-475.
Williams, D. R., & Williams, H. (1969). Auto-maintenance in the pigeon: Sustained pecking de-
spite contingent non-reinforcement. Journal of the Experimental Analysis of Behavior, 12, 511-520.
Wolpe, J. (1958). Psychotherapy by reciprocal inhibition. Stanford: Stanford University Press.
Yin, H., Barnet, R. C., & Miller, R. R. (1994). Second-order conditioning and Pavlovian condi-
tioned inhibition: Operational similarities and differences. Journal of Experimental Psychology: Animal
-303-
Behavior Processes, 20, 419-428.
Young, R. K. (1955). Retroactive inhibition and proactive inhibition under varying conditions of
response similarity. Journal of Experimental Psychology, 50, 113-119.
Received November 18, 2003.
Revision received April 4, 2004.
Accepted April 4, 2004.
... Described at length in the respondent conditioning literature (e.g., Escobar, Arcediano, & Miller, 2001a;Escobar, Matute, & Miller, 2001b;Matute & Pineño, Matute & Pineño, 1998;Miguez, Cham, & Miller, 2012;Miguez, Laborda, & Miller, 2014), most cases of contingency degradation also have been demonstrated in autoshaping preparations. Escobar and Miller (2004) described six specific ways in which CS-US contingencies are weakened by additional exposure to stimuli alone during nontraining trials. These appear in Fig. 3 of this article and are described in paragraphs that follow. ...
... Taken together, our understanding of CS-US contingency and its degradation indicates that practitioners should use forward pairing procedures (i.e., delay or trace conditioning) in which the target vocalization and preferred item occur on 100% of training trials and not without each other on training trials or outside of training trials. Although there are isolated demonstrations of facilitation (increased CR/ Schematic illustrating six types of contingency degradation reviewed by Escobar and Miller (2004). Exposure to vocalizations or preferred stimuli alone a before SSP training, b during SSP training, and c after SSP training can limit conditioning to practitioner vocalizations learning; e.g., potentiation) from extratrial exposure to the CS or US (e.g., Batsell & Batson, 1999;Honey & Hall, 1989), such additional, noncontingent/unpaired exposure typically hinders learning. ...
... One thing to note is that, although there exists a plethora of work on the impact of compound CSs (e.g., in blocking and overshadowing as described in Table 2), there is relatively little known about the use of compound USs during training (see Escobar & Miller, 2004). It is unknown whether overshadowing and competition between USs can occur in the same way as they occur for CSs. ...
Article
Full-text available
Stimulus–stimulus pairing (SSP) is a procedure used by behavior analysis practitioners that capitalizes on respondent conditioning processes to elicit vocalizations. These procedures usually are implemented only after other, more customary methods (e.g., standard echoic training via modeling) have been exhausted. Unfortunately, SSP itself has mixed research support, probably because certain as-yet-unidentified procedural variations are more effective than others. Even when SSP produces (or increases) vocalizations, its effects can be short-lived. Although specific features of SSP differ across published accounts, fundamental characteristics include presentation of a vocal stimulus proximal with presentation of a preferred item. In the present article, we draw parallels between SSP procedures and autoshaping, review factors shown to affect autoshaping, and interpret autoshaping research for suggested SSP tests and applications. We then call for extended use and reporting of SSP in behavior-analytic treatments. Finally, three bridges created by this article are identified: basic-applied, respondent–operant, and behavior analysis with other sciences.
... In Powers' (1973b) terms, they would correspond to disturbance and feedback function-updating mechanisms, respectively. 8 This error correction mechanism is known to be empirically instantiated in the negative accelerated shape in performance at the late stage of learning (e.g., Powers et al., 2012) and in stimulus competition phenomena, such as blocking and overshadowing 9 (see Escobar & Miller, 2004). One possible explanation of such a control mechanism is that it prevents boundless changes in the system from occurring, confining behavior within some nearoptimal margin (Bouton et al., 2020). ...
Article
Behavior can be regarded as the output of a system (action), as a function linking stimulus to response (reaction), or as an abstraction of the bidirectional relationship between the environment and the organism (interaction). When considering the latter possibility, a relevant question arises concerning how an organism can materially and continuously implement such a relationship during its lifetime in order to perpetuate itself. The feedback control approach has taken up the task of answering just that question. During the last several decades, said approach has been progressing and has started to be recognized as a paradigm shift, superseding certain canonical notions in mainstream behavior analysis, cognitive psychology, and even neuroscience. In this paper, we describe the main features of feedback control theory and its associated techniques, concentrating on its critiques of behavior analysis, as well as the commonalities they share. While some of feedback control theory's major critiques of behavior analysis arise from the fact that they focus on different levels of organization, we believe that some are legitimate and meaningful. Moreover, feedback control theory seems to blend with neurobiology more smoothly as compared to canonical behavior analysis, which only subsists in a scattered handful of fields. If this paradigm shift truly takes place, behavior analysts—whether they accept or reject this new currency—should be mindful of the basics of the feedback control approach.
... Besides possible semantic misattributions, the main argument supporting the idea that a single Pavlovian conditioning process operates in every Pavlovian conditioning procedure, is the fact that the same variables are critical for learning in all Pavlovian procedures (see Escobar & Miller, 2004 for reviews). The development of the CR relies mainly on the number of CS-US pairings, the temporal contiguity between the CS and the US, and the contingency between the CS and the US (the probability of the US in presence of the CS minus the probability of the US in its absence). ...
Article
Is Pavlovian conditioning the same thing as Pavlovian conditioning? Though that question seems tautological, this article shows that it is not, because Pavlovian conditioning has at least three different meanings: Pavlovian conditioning is (1) a procedure, (2) the learning phenomenon observed in that same procedure and (3) the learning process explaining the phenomenon observed in that procedure. If we look at this third meaning from an evolutionary point of view, it seems extremely unlikely that a single Pavlovian conditioning process is responsible for learning in all procedures classified as Pavlovian conditioning -- a conclusion that supported by behavioral and neural data. In the end, it seems that it might be better to drop the term Pavlovian conditioning to designate a learning process and to stop the quest for a single process explaining all Pavlovian learning. Instead, it would be more fruitful to understand under which condition a particular model of Pavlovian learning holds. The same conclusion applies to other research field in the psychology of learning, notably operant conditioning and statistical learning.
... En effet, à la suite des recherches pionnières menées par Pavlov et ses contemporains (1927), on assista à une étude systématique du conditionnement Pavlovien, notamment aux États-Unis avec le courant béhavioriste (Gormezano et Kehoe, 1981). Ces études permirent de mettre en évidence un ensemble de variables communes au conditionnement de comportements réflexes (Escobar et Miller, 2004). Ainsi il put par exemple être mis en évidence que la durée de l'intervalle entre le SC et le SI (e.g. ...
Thesis
Full-text available
Chez l’espèce humaine comme pour de nombreuses autres espèces animales, lorsque des stimuli environnementaux précèdent de façon régulière la présentation d’événements importants pour un individu, ces stimuli vont acquérir sous certaines conditions la capacité à évoquer des comportements dits d’anticipation. Cette capacité est considérée par de nombreux auteurs comme ayant une haute valeur adaptative, favorisant le contact avec des événements appétitifs et permettant l’évitement d’événements aversifs. Ces dernières décennies, deschercheurs ont initié un rapprochement entre le phénomène d’anticipation et le conditionnement Pavlovien. Ce rapprochement repose à la fois sur une similarité dans les caractéristiques des événements mis en jeux mais surtout sur de nombreux effets et phénomènes semblables, amenant ces auteurs à considérer que les comportements ditsd’anticipation, d’une façon générale, reposeraient sur le processus Pavlovien. Leconditionnement Pavlovien offre une littérature extrêmement riche dont l’une des principalesquestions de recherche concerne le problème des conditions à l’apparition du processus.Parmi les hypothèses existantes, l’Hypothèse de l’Information est sans aucun doute l’une desplus importantes par son influence. Selon cette hypothèse, un apprentissage associatifPavlovien n’aura lieu que lorsqu’un événement important sera présenté de façon inattendu à un sujet, et l’apprentissage, ou les associations apprises, ne porteront que sur des stimuli prédictifs de l’événement important (i.e. permettant son anticipation). A travers deux expériences appliquant une procédure de conditionnement rétrograde à une procédure de renforcement conditionné, nous avons cherché à tester les propositions faites par cette hypothèse. Nos résultats vont directement à l’encontre de ces propositions et vont au contraire dans le sens de deux autres propositions théoriques faites sur le conditionnement Pavlovien,illustrées par le modèle SOP et l’Hypothèse du Codage Temporel. Ces deux propositions sont testées au sein d’une troisième et dernière expérience, dont les implications pour ces modèles comme pour la conceptualisation du conditionnement Pavlovien et de l’anticipation de façon générale sont discutées.
... La literatura sobre reforzamiento condicionado sugiere que los principios involucrados en la adquisición de propiedades reforzantes de un estímulo pudieran ser paralelos a los principios involucrados en el establecimiento de estímulos condicionados en preparaciones de condicionamiento pavloviano (Dinsmoor, 1983). Un elemento común entre ambos es la proximidad temporal entre estímulos (Escobar & Miller, 2004;Fantino, 1977;Schneiderman & Gormezano, 1964;Smith, Coleman & Gormezano, 1969). En estudios de condicionamiento Pavloviano se ha observado que el condicionamiento de huella es generalmente menos eficaz para producir condicionamiento que el condicionamiento demorado. ...
... La literatura sobre reforzamiento condicionado sugiere que los principios involucrados en la adquisición de propiedades reforzantes de un estímulo pudieran ser paralelos a los principios involucrados en el establecimiento de estímulos condicionados en preparaciones de condicionamiento pavloviano (Dinsmoor, 1983). Un elemento común entre ambos es la proximidad temporal entre estímulos (Escobar & Miller, 2004;Fantino, 1977;Schneiderman & Gormezano, 1964;Smith, Coleman & Gormezano, 1969). En estudios de condicionamiento Pavloviano se ha observado que el condicionamiento de huella es generalmente menos eficaz para producir condicionamiento que el condicionamiento demorado. ...
Article
Full-text available
Without prior discrimination training, the acquisition of observing behavior could depend on accidental temporal proximity between the E+ and the primary reinforcer. Such temporal relations could be affected by varying stimuli duration. For evaluating the role of stimulus duration on the acquisition of observing responses, three groups of rats with no prior discrimination training were exposed to an observing-response procedure with 0.5, 5 and 10 s stimuli durations. It was found that observing behavior was established in a greater number of subjects with 10 s stimuli than with 5 or 0.5 s stimuli. These findings suggest that longer stimuli durations reduce the interval between the occurrence of the E+ and the primary reinforcer, facilitating the acquisition of observing responses.
... This outcome was obtained in situations in which associations were formed between the two CSs before both were allowed to predict the UCS. All these approaches imply that there is a single associative processor that determines which redundant cues gain ascendency (for reviews, see Escobar & Miller, 2004;R. A. Miller & Escobar, 2002). ...
Article
In 4 experiments, rats searched for food located on top of 4 of 16 towers which were arranged in a 4 × 4 matrix. The location of the baited towers was cued by visual landmark cues (the baited towers were striped, the others white) and by pattern cues (the baited towers were located in a 2 × 2 pattern within the larger 4 × 4 matrix) or simply by pattern cues without visual landmark cues. In 3 of the experiments, visual cues, after being paired with pattern cues, were removed altogether (Experiment 1), put into competition with pattern cues (Experiment 2), or made noninformative (Experiment 3). In Experiment 4, it was the pattern cues that were made noninformative. Collectively, the data suggest strongly that whereas the pattern is learned, even when presumably more salient visual cues are present, the connection between pattern and food location is much weaker than that between visual cue and food location. These data are more easily explained by a model of learning that includes dedicated modules than by a single-system associative model.
... A longstanding principle in Pavlovian conditioning is that cues are valid signals for other events to the extent that they reliably predict the occurrence of those events (e.g., Escobar and Miller, 2004). Logically, if a cue occurs in the absence of its predicted event or if the event occurs when the cue is absent, the predictive validity of that cue for the event will be diminished (e.g., Dwyer, Haselgrove, and Jones, 2011;Rescorla, 1968;Urushihara and Miller, 2009;Wagner, Logan, Haberlandt, and Price, 1968). ...
Article
An enormous amount of research has been aimed at identifying biological and environmental factors that are contributing to the current global obesity pandemic. The present paper reviews recent findings which suggest that obesity is attributable, at least in part, to a disruption of the Pavlovian control of energy regulation. Within our framework, this disruption occurs when (a) consumption of sweet-tasting, but low calorie or noncaloric, foods and beverages reduces the ability of sweet tastes to predict the postingestive caloric consequences of intake and (b) consuming diets high in saturated fat and sugar (a.k.a., Western diet) impairs hippocampal-dependent learning and memory processes that are involved with the use of interoceptive "satiety" signals to anticipate when food and eating are not followed by appetitive postingestive outcomes. The paper concludes with discussion of a "vicious-cycle' model which links obesity to cognitive decline.
Chapter
Evolution of Learning and Memory Mechanisms is an exploration of laboratory and field research on the many ways that evolution has influenced learning and memory processes, such as associative learning, social learning, and spatial, working, and episodic memory systems. This volume features research by both outstanding early-career scientists as well as familiar luminaries in the field. Learning and memory in a broad range of animals are explored, including numerous species of invertebrates (insects, worms, sea hares), as well as fish, amphibians, birds, rodents, bears, and human and nonhuman primates. Contributors discuss how the behavioral, cognitive, and neural mechanisms underlying learning and memory have been influenced by evolutionary pressures. They also draw connections between learning and memory and the specific selective factors that shaped their evolution. Evolution of Learning and Memory Mechanisms should be a valuable resource for those working in the areas of experimental and comparative psychology, comparative cognition, brain–behavior evolution, and animal behavior.
Article
Orthodox psychiatric texts are often rich in facts, but thin in concept. Depression may be defined as a dysfunction of mood, but of what use is a mood? How can anxiety be both symptom and adaptation to stress? What links the disparate disabilities of perception and reasoning in schizophrenia? Why does the same situation push one person into drink, drugs, danger, or despair and bounce harmlessly off another? Trouble in Mind is unorthodox because it models adaptive mental function along with mental illness to answer questions like these. From experience as a Johns Hopkins clinician, educator, and researcher, Dean F. MacKinnon offers a unique perspective on the nature of human anguish, unreason, disability, and self-destruction. He shows what mental illness can teach about the mind, from molecules to memory to motivation to meaning. MacKinnon's fascinating model of the mind as a vital function will enlighten anyone intrigued by the mysteries of thought, feeling, and behavior. Clinicians in training will especially appreciate the way mental illness can illuminate normal mental processes, as medical illness in general teaches about normal body functions. For students, the book also includes useful guides to psychiatric assessment and diagnosis. © 2011 The Johns Hopkins University Press. All rights reserved.
Article
(This partially reprinted article originally appeared in Psychological Review, 1950, Vol 57, 94–207. The following abstract of the original article appeared in PA, Vol 24:5093.) An attempt has been made to clarify some issues in current learning theory by giving a statistical interpretation to the concepts of stimulus and response and by deriving quantitative laws that govern simple behavior systems. Dependent variables, in this formulation, are classes of behavior samples with common quantitative properties; independent variables are statistical distributions of environmental events. Laws of the theory state probability relations between momentary changes in behavioral and environmental variables. From this point of view it has been possible to derive simple relations between probability of response and several commonly used measures of learning, and to develop mathematical expressions describing learning in both classical conditioning and instrumental learning situations under simplified conditions.
Article
Four conditioned suppression experiments examined the influence of contextual stimuli on the rat's fear of an extinguished conditioned stimulus (CS). When rats received pairings of a CS with shock in one context and then extinction of the CS in another context, fear of the CS was renewed when the CS was returned to and tested in the original context (Experiments 1 and 3). No such renewal was obtained when the CS was tested in a second context after extinction had occurred in the conditioning context (Experiment 4). In Experiment 2, shocks presented following extinction reinstated fear of the CS, but only if they were presented in the context in which the CS was tested. In each experiment, the associative properties of the contexts were independently assessed. Contextual excitation was assessed primarily with context-preference tests in which the rats chose to sit in either the target context or an adjoining side compartment. Contextual inhibition was assessed with summation tests. Although reinstatement was correlated with demonstrable contextual excitation present during testing, the renewal effect was not. Moreover, there was no evidence that contextual inhibition developed during extinction. The results suggest that fear of an extinguished CS can be affected by the excitatory strength of the context but that independently demonstrable contextual excitation or inhibition is not necessary for contexts to control that fear.
Article
The aim of chemotherapeutic tumor therapy is the complete elimination of malignant cells. Chemotherapeutic drugs inhibit cell proliferation, i. e., they have cytostatic qualities. But they also lead to cytolysis, thus having cytotoxic properties. Patients receiving chemotherapy often experience nausea, vomiting, appetite loss, and immunomodulation. These side-effects not only occur after delivery of the drugs. Patients also complain about nausea, vomiting, and appetite loss prior to a chemotherapy-session. There are also preliminary data showing anticipatory immunomodulation. The following paper reviews these behavioral side-effects with the emphasis on nausea and vomiting. In addition to previously published papers, the relationship between the different conditioned side-effects (anticipatory nausea and vomiting [ANV], learned food aversion [LFA], and anticipatory immunomodulation [AIM]) is analyzed, and results on conditioned emotional distress are summarized. Possible physiological and endocrine mediators of unconditioned and conditioned side-effects are reviewed and research strategies for further studies are discussed.
Chapter
This chapter describes the potential explanatory power of a specific response rule and its implications for models of acquisition. This response rule is called the “comparator hypothesis.” It was originally inspired by Rescorla's contingency theory. Rescorla noted that if the number and frequency of conditioned stimulus–unconditioned stimulus (CS–US) pairings are held constant, unsignaled presentations of the US during training attenuate conditioned responding. This observation complemented the long recognized fact that the delivery of nonreinforced presentations of the CS during training also attenuates conditioned responding. The symmetry of the two findings prompted Rescorla to propose that during training, subjects inferred both the probability of the US in the presence of the CS and the probability of the US in the absence of the CS and they then established a CS–US association based upon a comparison of these quantities. The comparator hypothesis is a qualitative response rule, which, in principle, can complement any model of acquisition.