ArticlePDF Available

Abstract and Figures

It is generally assumed that the function of contingency learning is to predict the occurrence of important events in order to prepare for them. This assumption, however, has scarcely been tested. Moreover, the little evidence that is available suggests just the opposite result. People do not use contingency to prepare for outcomes, nor to predict their occurrence, although they do use it to infer the causal and predictive value of cues. By using both judgmental and behavioral data, we designed the present experiments as a further test for this assumption. The results show that-at least under certain conditions-people do use contingency to prepare for outcomes, even though they would still not use it to predict their occurrence. The functional and adaptive aspects of these results are discussed in the present article.
Content may be subject to copyright.
It is generally assumed that in order to survive, people
and animals need to extract contingency information from
the environment. Doing this allows them to infer causal
relations and to predict when important events will occur,
which in turn allows them to prepare for these outcomes
(see, e.g., Dickinson, 1980; Hollis, 1997). The large num-
ber of experiments on contingency learning that have been
published during the last decades (see Shanks, 2007, for
a recent review) are justified precisely on these premises.
Contingency learning is critical for survival because it ul-
timately allows organisms to prepare for important events.
A functional analysis of contingency learning, therefore,
calls for the investigation of preparatory behavior. Thus,
the scant attention that preparatory judgments have cap-
tured in the research community is highly surprising. Pre-
paratory judgments are not just one more additional type
of judgment. They are the judgments that most naturally
show the function that has traditionally been attributed to
contingency learning. Indeed, if preparing for outcomes
provides the function of contingency learning, then one
would expect preparatory behavior to be closely depen-
dent on contingency information. Quite surprisingly, the
(little) data available on preparatory judgments suggest
otherwise.
The most commonly used metric of contingency is ΔP,
which is the difference between the probability of the out-
come when the cue is present, p(O|C), and the probability
of the outcome when the cue is absent, p(O|¬C) (Jenkins
& Ward, 1965). The experimenter can manipulate the
strength of the cueoutcome contingency by varying the
frequencies with which the cue (e.g., a fictitious medi-
cine) and the outcome (e.g., a fictitious allergic reaction)
occur or do not occur together. Interestingly, some reports
have suggested that these contingency manipulations
have a different effect on causal judgments than on pre-
dictive and preparatory judgments. For example, Vadillo,
Miller, and Matute, (2005; see also Vadillo & Matute,
2007) showed that contingency information had no im-
pact on participants’ likelihood of predicting the outcome
in the presence of the cue. At the same time, however,
participants in those studies showed an accurate use of
contingency information to estimate both the causal and
the predictive value of the cue. Thus, apparently, people
use contingency to infer causal and predictive value, even
though they do not use it to predict the likelihood that the
outcome will follow.
In a related study, De Houwer, Vandorpe, and Beckers
(2007) showed that contingency information was used to
make accurate estimates of causal relations, whereas it
had little impact on participants’ stating that they would
prepare for the outcome when the cue was present. Con-
sistent with the findings of Vadillo et al. (2005) regarding
outcome predictions, p(O|C) seemed to be the main deter-
minant of preparatory judgments as well.
At least at first glance, those results seem quite con-
trary to the idea that opened the present articlethat
contingency knowledge is needed both to predict the oc-
currence of important events and to prepare for them.
From a normative point of view, it certainly does make
sense to neglect the value of contingency when making
predictions. After all, the value of p(O|C) is the only
thing one needs to know when predicting whether the
outcome will follow the cue. Information on p(O|¬C)
becomes absolutely irrelevant in this case. But the poten-
tial dissociation between contingency and preparatory
behavior is not that obvious. On the one hand, prepara-
tory judgments, like predictions, could simply rely on
p(O|C). People could prepare for the outcome whenever
117 © 2010 The Psychonomic Society, Inc.
Contingency is used to prepare for outcomes:
Implications for a functional analysis of learning
Fe r n a n d o Bl a n c o , He l e n a Ma t u t e , a n d Mi g u e l a. V a d i l l o
Deusto University, Bilbao, Spain
It is generally assumed that the function of contingency learning is to predict the occurrence of important
events in order to prepare for them. This assumption, however, has scarcely been tested. Moreover, the little evi-
dence that is available suggests just the opposite result. People do not use contingency to prepare for outcomes,
nor to predict their occurrence, although they do use it to infer the causal and predictive value of cues. By using
both judgmental and behavioral data, we designed the present experiments as a further test for this assumption.
The results show that—at least under certain conditions—people do use contingency to prepare for outcomes,
even though they would still not use it to predict their occurrence. The functional and adaptive aspects of these
results are discussed in the present article.
Psychonomic Bulletin & Review
2010, 17 (1), 117-121
doi:10.3758/PBR.17.1.117
H. Matute, matute@fice.deusto.es
118 Bl a n c o , Mat u t e , a n d Va d i l l o
has not taken Dugetil and a picture of a bottle crossed in red were
shown. To encourage their attention, participants had to answer the
question Do you think that Mr. X is going to develop a skin rash?
by clicking the “yes” or “no” button in every trial. Then, a feedback
panel showed either the phrase Mr. X has developed the skin rash
and a picture of an ill face (outcome-present trials), or the phrase
Mr. X has not developed the skin rash and a picture of a healthy face
(outcome-absent trials).
At test, every participant emitted three different judgments:
a causal judgment (To what extent do you think that Dugetil is the
cause of the skin rash that Mr. X suffers?), a prediction judgment
(If Mr. X has taken Dugetil, to what extent do you think that he will
develop a skin rash?), and a preparation judgment (If Mr. X has taken
Dugetil, to what extent would you recommend him to take an oint-
ment in order to prevent a skin rash?). The three questions were dis-
played simultaneously, with their position (upper, middle, or lower
panel of the screen) counterbalanced between participants. A rating
scale ranging from 0 (labeled Definitely NOT ) to 100 (labeled Defi-
nitely YES) was displayed below each question.
Results
Figure 1 depicts the mean judgments at test. As was ex-
pected, a 3 (question) 3 2 (contingency) ANOVA showed
significant main effects of question [F(2,98) 5 27.10, p ,
.001] and contingency [F(1,49) 5 17.07, p , .001], as well
as an interaction [F(2,98) 5 6.14, p , .01]. Both causal
and preparatory judgments were significantly higher in
Group High than in Group Low [t(49) 5 6.46, p , .001,
and t(49) 5 2.41, p , 0.05, respectively]. This indicates
that both of these judgments were affected by contingency.
Prediction judgments did not significantly differ as a func-
tion of contingency [t(49) 5 1.28, p 5 .21].
EXPERIMENT 2
As was expected, in Experiment 1 we showed that both
causal and preparatory judgments, but not predictions,
were affected by contingency information. The results of
the causal and prediction judgments were consistent with
those of several previous human studies (e.g., Gredebäck
et al., 2000; Matute et al., 2002; Vadillo et al., 2005; Va-
dillo & Matute, 2007). However, the results on prepara-
tory judgments, although expected, contradicted the null
they know it will occur. On the other hand, however,
it is not clear that the most adaptive behavior that one
can adopt when preparing for an outcome should rely
simply on p(O|C). If the outcome is occurring with the
same probability regardless of whether the target cue is
present or absent, this becomes a situation in which the
outcome follows the cue regularly, but without any pre-
dictive relation between them. The cue is not a signal for
the outcome, but a completely unrelated event. In this
situation, organisms should look for another cue with
greater predictive value to help them decide when to act,
or else they will have to be preparing at all times for any
of the many possible outcomes that could follow any of
the millions of cues that are present at any given time.
There are so many potential outcomes for which there
are no known causes (or good signals) that it becomes
impossible to prepare for them all. Thus, when prepar-
ing for outcomes, it might be much more efficient to
rely on the predictive value of the cue, which depends on
contingency, rather than on p(O|C). Indeed, if any con-
clusion can be drawn from decades of related research
on Pavlovian conditioning, it is that nonhuman animals
do use contingency to prepare for outcomes (Rescorla,
1968). As for human preparatory Pavlovian responses,
we know of no study that has tested whether they rely on
cueoutcome contingency rather than on p(O|C).
In sum, further evidence is needed before we can reject
(or accept) the general assumption that people use con-
tingency information to prepare for outcomes. Although
several studies have shown a dissociation between causal
judgments and predictions (e.g., Gredeck, Winman,
& Juslin, 2000; Matute, Vegas, & De Marez, 2002; Va-
dillo et al., 2005; Vadillo & Matute, 2007), the evidence
of the dissociation between contingency and preparatory
judgments relies on only one study (De Houwer et al.,
2007). Moreover, none of the aforementioned studies has
explored the effects of contingency information on both
outcome predictions and preparatory judgments. If our
hypothesis is correct, then preparatory judgments, not
predictions, might show sensitivity to contingency.
EXPERIMENT 1
Method
Participants
. Fifty-one college students voluntarily took part in
the study. A random assignment of participants into two groups re-
sulted in 24 participants in Group High, and 27 in Group Low.
Procedure and Design
. Participants were told to imagine that
they were physicians who were trying to find out whether a relation-
ship existed between a cue (a fictitious medicine, Dugetil) and an
outcome (a side effect—skin rash—in a fictitious patient).
p(O|C) was 1 in both groups, but p(O|¬C) was .5 in Group High
and 1 in Group Low. Therefore, Δp was .5 in Group High and 0
in Group Low. This should permit us to compare the sensitivity of
causal, prediction, and preparation judgments to specific manipula-
tions of covariational information. Judgments that are only sensitive
to p(O|C) should not differ between these two groups, but judgments
that are sensitive to Δp should.
Participants received 120 medical cards, 1 per trial. In cue- present
trials, the computer screen showed the phrase Mr. X has taken Dug-
etil and a picture of a pill bottle. In cue-absent trials, the phrase Mr. X
0
25
50
75
100
Causal Prediction Preparatory
Mean Judgments
Type of Question
Group High
Group Low
Figure 1. Mean judgments in Experiment 1. Error bars repre-
sent 1 standard error of the mean.
co n t i n g e n c y , Pr e d i c t i o n , a n d Pr e P a r a t i o n 119
spaceships invade the screen faster than usual (one spaceship every
50 msec). These invasions cannot be escaped once started. Thus, to
avoid them, participants have to suppress their barpressing behavior
immediately before the shield is connected. Participants are also
told that there will be some cues that sometimes may (or may not)
be followed by the activation of the shield. If participants learn that
a cue is a good predictor of the shield, then they should respond to
this cue by suppressing their barpressing behavior in anticipation
of the outcome. Suppression ratios are conventionally assessed as
A/ (A 1 B), where A is the number of barpresses during the cue and
B the number of barpresses in a period of time identical to the dura-
tion of the cue and immediately preceding it.
In Experiment 2, the training phase consisted of 69 trials in
pseudo random order. In both groups, the target cue was presented
in 30 trials, and the activation of the shield followed 80% of them
(24 trials). Contingency was manipulated by varying the percentage
of trials in which the context, rather than the cue, was followed by
the outcome. In the high-contingency condition, the context was
never followed by the activation of the shield, whereas in the low-
contingency condition, the context was followed by the shield 80%
of the time (i.e., 24 trials), thereby making the target cue noncontin-
gent on the shields despite its high probability of being reinforced.
An additional 15 trials were used in both groups with a filler cue
that was never followed by the shield, and its only purpose was to
prevent indiscriminate suppression (Arcediano et al., 1996). The
target and filler cues were two “Martian letters,counterbalanced,
that represented the interception of messages between spaceships
(Costa, 2009). They appeared from time to time for a duration of
1.5 sec. At test, the target cue was presented for 3 sec so that the
suppression ratio could be assessed (see Arcediano et al., 1996).
Intertrial intervals lasted between 7 and 13.50 sec.
Upon completion of the Martians task, the judgmental variables
were assessed between participants. The screen presented the target
cue and a question that had to be answered with a number between
0 and 100. Half of the participants in each contingency condition re-
ceived a prediction question: If this cue [image of target cue inserted
here] appears on the screen, to what extent do you think that the
shield will be activated? The other half received a predictive-value
question: To what extent do you think that the onset of this cue is a
good predictor of the activation of the shield?
Results
Behavioral data
. The results of Experiment 2 were as
expected. Participants in the high-contingency condition
suppressed their barpressing behavior more in response
to the target cue (M 5 .20; SEM 5 .11) than did those
in the low-contingency condition (M 5 .27; SEM 5 .23)
[F(1,32) 5 7.08, p , .05]. (Recall that values closer to 0
indicate stronger suppression—that is, stronger prepara-
tory behavior in response to the target cue.) Thus, partici-
pants’ preparatory behavior was significantly affected by
the contingency between the cue and the shield, with those
exposed to a higher contingency being the ones who sup-
pressed their behavior the most.
Judgments
. Figure 2 plots the mean judgments by
group in this experiment. A 2 (contingency) 3 2 (question)
ANOVA yielded a significant interaction [F(1,30) 5 6.46,
p , .05]. No main effects for contingency [F(1,30) 5 2.40,
p 5 .132] or for question [F(1,30) 5 2.20, p 5 .148] were
found. Consistent with our hypothesis and with previous
literature, predictive-value judgments were significantly
affected by the contingency manipulation [F(1,16) 5
10.51, p , .01], whereas prediction judgments were not
[F(1,14) 5 3.97, p 5 . 54].
result reported by De Houwer et al. (2007). Our Experi-
ment 2, therefore, had several goals. The first one was
to provide a replication for the results of Experiment 1
using a very different procedure, in which most param-
eters changed, allowing the generality of the effect to be
assessed. The most important change in this experiment
was probably that in our dependent variable. A functional
analysis of contingency learning cannot rely solely on
what people say that they would do (i.e., their subjective,
verbalized judgments), or what they say that somebody
else should do. Our analysis needed convergent evidence
from the actual, nonverbal, preparatory behavior. Experi-
ment 2 was aimed at extending the findings of Experi-
ment 1 to a more comprehensive framework that included
not only verbal (numerical) judgments but also, most
importantly, the nonverbal preparatory behavior itself.
On a lesser point, we also aimed at using less extreme
contingencies. Although extreme values were needed
for a fair comparison with previous research [i.e., both
De Houwer et al., 2007, and Vadillo et al., 2005, used
a contingency of 0 and a p(O|C) of 1], and although the
difference between the contingency and p(O|C) needed to
be large for the dissociation to be observed, we thought
it was important to use a lower p(O|C) so as to make sure
that the results were not restricted to the special condition
in which the outcome is always present. Also, in order
to make sure that the dissociation observed in the judg-
mental responses in Experiment 1 was not the result of
our presenting all of the questions in the same screen,
we manipulated the question type between participants,
with each participant answering only one question. Fi-
nally, we used predictive-value rather than causal-value
questions. If our hypothesis was correct, the dissociation
that we observed between prediction and causal-value
judgments should also have been observed between pre-
diction and predictive-value judgments (see, e.g., Vadillo
et al., 2005).
Method
Participants
. Thirty-four anonymous volunteers took part in the
study. A random assignment resulted in 16 participants in the high-
contingency condition, and 18 in the low-contingency condition.
Orthogonally, 16 participants emitted a prediction judgment at the
end of training, and 18 emitted a prediction-value judgment.
Procedure and Design
. In Experiment 2, we used the Martians
video game that was originally developed by Arcediano, Ortega,
and Matute (1996); however, we used the new and improved version
that was developed by Franssen, Clarysse, Beckers, van Vooren, and
Baeyens (in press),
1
and we also incorporated several interesting
features that were developed by Costa (2009; see note 1). In the Mar-
tians paradigm, the suppression of barpressing behavior is used as an
analogue to animal conditioning suppression to assess preparatory
behavior in humans. The participants’ goal is to prevent Martians
from landing on Earth. Martian spaceships appear on the computer
screen at a rate of one spaceship every 250 msec. Participants are
told to fire their laser guns by pressing the space bar, and to destroy
as many spaceships as possible. Doing this results in a stable bar-
pressing rate. From time to time, the Martians activate an anti laser
shield consisting of screen flashes and a distinctive sound with a
duration of 500 msec. This shield is the outcome to be prepared
for. If the participant continues pressing the bar when the shield is
active, the laser gun becomes deactivated for 5 sec, and dozens of
120 Bl a n c o , Mat u t e , a n d Va d i l l o
We cannot know the reason why De Houwer et al.
(2007) observed a null result in the sensitivity of prepara-
tory judgments to contingency, but the main differences
between our judgmental study (Experiment 1) and theirs
are that we manipulated contingency between, rather than
within, participants; we used 120 training trials rather
than 10; and we assessed preparatory responses differ-
ently. These, and perhaps several other differences we are
not aware of, increased the chances that the sensitivity of
preparatory judgments to contingency could be detected
in our studies.
Why, then, is contingency learning used to prepare for
outcomes if it is not used when predicting their occurrence?
If the outcome is occurring often, regardless of whether the
cue is present or absent, a normatively adjusted response
should be sensitive to several facts: (1) The contingency
between the cue and the outcome is 0; (2) the causal value
of the cue is 0; (3) the predictive value of the cue is 0; (4) the
probability that the outcome will occur in the next trial is
high; and (5) the likelihood (the prediction) that the out-
come will occur following the cue is high. However, which
would be the most effective preparatory behavior? A prepa-
ratory response speaks to the question of what should be the
best possible use of knowledge and information in order to
best adapt to the demands of the environment. This includes
making good use of knowledge and information, but also
of energy resources and cost-efficient behavior. Therefore,
this also includes assessing the high cost of emitting a pre-
paratory response in all trials, irrespective of the predictive
value of the cue. What the present experiments have shown
is that once participants know that the outcome will occur
with a given probability and that there are no clues as to
what its cause can be, they learn that it makes no sense to
keep preparing for that outcome whenever any of the many
irrelevant cues in the environment are present (or at least it
makes no sense to prepare as intensely as if it were contin-
gent). Although we are often able to predict the occurrence
of important events just by looking at their base rates, it
would be impossible to prepare for all types of outcomes
that may occur, irrespective of their degree of contingency
with the signals that are present.
Because predictions, preparatory behavior, and prediction-
value judgments are so often taken as synonyms, we all tend
to assume that animals use contingency to predict the out-
comes. However, the only thing we can know for sure when
we work with animals is that they use contingency to prepare
for outcomes. Whether they use contingency because they
predict the outcome or because they know that the cue has
a high predictive value is something that, to our knowledge,
has not been addressed in the animal literature. Our results
suggest that, in addition to preparation behavior, it is the pre-
dictive value of cues rather than the predictions of the out-
come that requires contingency. This point should be valid
for human and nonhuman animals.
In sum, the present experiments have shown that con-
tingency learning may have survival value after all. It is
true that we do not need it to know the likelihood that
an outcome will occur, but we do need it if we want to
be able to select those signals with the higher predictive
(or causal) value in our environment so that we can en-
DISCUSSION
As was noted in the introduction, it is often assumed
that the reason why humans and animals extract contin-
gency information from the environment is a very practi-
cal one: being able to identify those cues that signal im-
portant events, so as to be able to predict their occurrence
and prepare for them.
Previous research has indeed shown that animals use
contingency information to prepare for outcomes (Res-
corla, 1968). It has also shown that people use contin-
gency information to acquire knowledge not only about
causation, but also about the predictive (signal) value of
the cues in the environment (Vadillo et al., 2005; Vadillo
& Matute, 2007). However, it has also shown that people
do not use this knowledge to predict the outcomes (see,
e.g., Gredebäck et al., 2000; Matute et al., 2002; Vadillo
et al., 2005; Vadillo & Matute, 2007) or to prepare for
them (De Houwer et al., 2007). This seems inconsistent
with the general assumption that the function of contin-
gency learning is to predict and prepare for outcomes.
Our present experiments have shown that prepara-
tory judgment and behavior are sensitive to contingency.
The two experiments used very different procedures and
dependent variables and demonstrated consistently that
preparatory judgments and behavior were significantly
stronger when there was a high level of contingency. Par-
ticipants also used contingency to detect which cues had
the higher causal and predictive values. Consistent with
previous reports, however, outcome predictions were not
based on contingency (see, e.g., Gredebäck et al., 2000;
Matute et al., 2002; Vadillo et al., 2005; Vadillo & Matute,
2007). Similar dissociations have been found for judg-
ments about the frequency with which the outcome fol-
lows the cue (Matute, Arcediano, & Miller, 1996) and
for outcome recall in the presence of the cue (Mitchell,
Lovibond, & Gan, 2005). Both of them also appear to
be independent of contingency and different from causal
judgments. Moreover, related research is also showing
that people tend to use base rates, rather than contingen-
cies, to infer correlations and to make predictions (Fiedler,
Freytag, & Meiser, 2009).
0
25
50
75
100
Prediction Predictive Value
Mean Judgments
Type of Question
Group High
Group Low
Figure 2. Mean judgments in Experiment 2. Error bars repre-
sent 1 standard error of the mean.
co n t i n g e n c y , Pr e d i c t i o n , a n d Pr e P a r a t i o n 121
Franssen, M., Clarysse, J., Beckers, T., van Vooren, P. R., & Baey-
ens, F. (in press). A free software package for a human online condi-
tioned suppression preparation. Behavior Research Methods.
Gredebäck, G., Winman, A., & Juslin, P. (2000). Rational assessments
of covariation and causality. In L. R. Gleitman & K. Joshi (Eds.), Pro-
ceedings of the 22nd Annual Conference of the Cognitive Science So-
ciety (pp. 190-195). Mahwah, NJ: Erlbaum.
Hollis, K. L. (1997). Contemporary research on Pavlovian condition-
ing: A “new” functional analysis. American Psychologist, 52, 956-965.
doi:10.1037/0003-066X.52.9.956
Jenkins, H. M., & Ward, W. C. (1965). Judgment of contingency be-
tween responses and outcomes. Psychological Monographs, 79, 1-17.
Matute, H., Arcediano, F., & Miller, R. R. (1996). Test question
modulates cue competition between causes and between effects. Jour-
nal of Experimental Psychology: Learning, Memory, & Cognition, 22,
182-196. doi:10.1037/0278-7393.22.1.182
Matute, H., Vegas, S., & De Marez, P. J. (2002). Flexible use of recent
information in causal and predictive judgments. Journal of Experi-
mental Psychology: Learning, Memory, & Cognition, 28, 714-725.
doi:10.1037/0278-7393.28.4.714
Mitchell, C. J., Lovibond, P. F., & Gan, C. Y. (2005). A dissociation
between causal judgment and outcome recall. Psychonomic Bulletin
& Review, 12, 950-954.
Rescorla, R. A. (1968). Probability of shock in the presence and ab-
sence of CS in fear conditioning. Journal of Comparative & Physi-
ological Psychology, 66, 1-5. doi:10.1037/h0025984
Shanks, D. R. (2007). Associationism and cognition: Human contin-
gency learning at 25. Quarterly Journal of Experimental Psychology,
60, 291-309. doi:10.1080/17470210601000581
Vadillo, M. A., & Matute, H. (2007). Predictions and causal es-
timations are not supported by the same associative structure.
Quarterly Journal of Experimental Psychology, 60, 433-447.
doi:10.1080/17470210601002520
Vadillo, M. A., Miller, R. R., & Matute, H. (2005). Causal and
predictive-value judgments, but not predictions, are based on cue
outcome contingency. Learning & Behavior, 33, 172-183.
NOTE
1. We thank these researchers for sending us their improved versions
of the program. Special thanks are due to M. Franssen for helping us
program the experiment using the Leuven version.
(Manuscript received June 3, 2009;
revision accepted for publication October 9, 2009.)
gage in efficient preparatory behavior. Most importantly,
and regardless of the merits of the particular account we
have presented herein, our research calls for the need to
add preparatory judgments and behavior to the research
agenda of contingency learning. As was noted in the intro-
duction, the little attention that preparatory judgments and
behavior have captured in the human learning research
tradition is surprising, and it will be a task for future re-
search to investigate under which conditions contingency
learning is or is not useful for preparing for outcomes.
AUTHOR NOTE
Support for the present research was provided by Grant SEJ2007-63691/
PSIC from Dirección General de Investigación of the Spanish Govern-
ment, Grant PI2008-9 from Departamento de Educación, Universidades e
Investigación of the Basque Government, and Grant P08-SEJ-3586 from
Junta de Andalucía. F.B. was supported by fellowships BFI081.020.0 and
BFI04.484 from the Basque Government. He is now at the University
of Leuven, Belgium. We would like to thank M. Brown, J. De Houwer,
S. Musca, and C. Orgaz for their insightful comments on preliminary ver-
sions of this manuscript. Special thanks are due to Estíbaliz San Antón
for her valuable help in recruiting and running the participants of Experi-
ment 2. Correspondence concerning this article should be addressed to
H. Matute, Departamento de Psicología, Universidad de Deusto, Apar-
tado 1, 48080 Bilbao, Spain (e-mail: matute@deusto.es).
REFERENCES
Arcediano, F., Ortega, N., & Matute, H. (1996). A behavioral prepara-
tion for the study of human Pavlovian conditioning. Quarterly Journal
of Experimental Psychology, 49B, 270-283. doi:10.1080/713932633
Costa, D. S. J. (2009). Maintenance of behaviour when reinforcement
becomes delayed. Unpublished doctoral dissertation, University of
Sydney.
De Houwer, J., Vandorpe, S., & Beckers, T. (2007). Statistical contin-
gency has a different impact on preparation judgments than on causal
judgments. Quarterly Journal of Experimental Psychology, 60, 418-
432. doi:10.1080/17470210601001084
Dickinson, A. (1980). Contemporary animal learning theory. Cam-
bridge: Cambridge University Press.
Fiedler, K., Freytag, P., & Meiser, T. (2009). Pseudocontingencies:
An integrative account of an intriguing cognitive illusion. Psychologi-
cal Review, 116, 187-206. doi:10.1037/a0014480
... Most causal learning experiments exploit a basic principle of causality: causes and effects (outcomes) correlate with each other, unless a third factor masks this relationship. Since causality cannot be directly observed (Hume, 1748), people use this simple principle and rely on a proxy measure, the contingency between the cause and the outcome, to estimate causality (Allan, 1980;Wasserman et al., 1996;Vadillo et al., 2005;Blanco et al., 2010). In a simple situation with only one binary cause and one binary outcome, the contingency can be computed by means of the p index (Allan, 1980). ...
... Previous experiments that used causal learning paradigms suggest that people can often be accurate in their judgments of causality (Shanks and Dickinson, 1987;Wasserman, 1990;Blanco et al., 2010), being generally sensitive to the actual contingency presented in the experiments. However, researchers have also reported systematic deviations, or biases. ...
Article
Causal illusions have been postulated as cognitive mediators of pseudoscientific beliefs, which, in turn, might lead to the use of pseudomedicines. However, while the laboratory tasks aimed to explore causal illusions typically present participants with information regarding the consequences of administering a fictitious treatment versus not administering any treatment, real-life decisions frequently involve choosing between several alternative treatments. In order to mimic these realistic conditions, participants in two experiments received information regarding the rate of recovery when each of two different fictitious remedies were administered. The fictitious remedy that was more frequently administered was given higher effectiveness ratings than the low-frequency one, independent of the absence or presence of information about the spontaneous recovery rate. Crucially, we also introduced a novel dependent variable that involved imagining new occasions in which the ailment was present and asking participants to decide which treatment they would opt for. The inclusion of information about the base rate of recovery significantly influenced participants’ choices. These results imply that the mere prevalence of popular treatments might make them seem particularly effective. It also suggests that effectiveness ratings should be interpreted with caution as they might not accurately reflect real treatment choices. Materials and datasets are available at the Open Science Framework [https://osf.io/fctjs/].
... Most causal learning experiments exploit a basic principle of causality: causes and effects (outcomes) correlate with each other, unless a third factor masks this relationship. Since causality cannot be directly observed (Hume, 1748), people use this simple principle and rely on a proxy measure, the contingency between the cause and the outcome, to estimate causality (Allan, 1980;Wasserman et al., 1996;Vadillo et al., 2005;Blanco et al., 2010). In a simple situation with only one binary cause and one binary outcome, the contingency can be computed by means of the p index (Allan, 1980). ...
... Previous experiments that used causal learning paradigms suggest that people can often be accurate in their judgments of causality (Shanks and Dickinson, 1987;Wasserman, 1990;Blanco et al., 2010), being generally sensitive to the actual contingency presented in the experiments. However, researchers have also reported systematic deviations, or biases. ...
Article
Full-text available
Patients' beliefs about the effectiveness of their treatments are key to the success of any intervention. However, since these beliefs are usually formed by sequentially accumulating evidence in the form of the covariation between the treatment use and the symptoms, it is not always easy to detect when a treatment is actually working. In Experiments 1 and 2, we presented participants with a contingency learning task in which a fictitious treatment was actually effective to reduce the symptoms of fictitious patients. However, the base-rate of the symptoms was manipulated so that, for half of participants, the symptoms were very frequent before the treatment, whereas for the rest of participants, the symptoms were less frequently observed. Although the treatment was equally effective in all cases according to the objective contingency between the treatment and healings, the participants' beliefs on the effectiveness of the treatment were influenced by the base-rate of the symptoms, so that those who observed frequent symptoms before the treatment tended to produce lower judgments of effectiveness. Experiment 3 showed that participants were probably basing their judgments on an estimate of effectiveness relative to the symptom base-rate, rather than on contingency in absolute terms. Data and materials are publicly available at the Open Science Framework: https://osf.io/emzbj/
... Thus, as causes and effects are expected to correlate, detecting these relationships, or contingencies, is a valuable skill to infer potential causality. In line with this reasoning, a great corpus of experimental evidence strongly suggests that manipulations of the contingency between potential causes and effects produce the corresponding variations in people's judgments of causality (Allan and Jenkins, 1980;Shanks and Dickinson, 1987;Dodwell and Humphrey, 1993;Blanco et al., 2010). ...
Article
Full-text available
The causal illusion is a cognitive bias that results in the perception of causality where there is no supporting evidence. We show that people selectively exhibit the bias, especially in those situations where it favors their current worldview as revealed by their political orientation. In our two experiments (one conducted in Spain and one conducted in the United Kingdom), participants who self-positioned themselves on the ideological left formed the illusion that a left-wing ruling party was more successful in improving city indicators than a right-wing party, while participants on the ideological right tended to show the opposite pattern. In sum, despite the fact that the same information was presented to all participants, people developed the causal illusion bias selectively, providing very different interpretations that aligned with their previous attitudes. This result occurs in situations where participants inspect the relationship between the government’s actions and positive outcomes (improving city indicators) but not when the outcomes are negative (worsening city indicators).
... Contingency refers to the extent to which the potential cause and the alleged consequence co-vary (in the previous example, the extent to which studying and scoring a good mark correlate), and it is usually measured with the Δp index, computed as the difference between the probability of the consequence in the presence of the potential cause and in its absence [1,2]. A large number of studies have shown that humans are sensitive to the contingency between events [3][4][5][6], and there is also some evidence showing that even very young children may be able to infer causality by using co-variation patterns [7]. ...
Article
Full-text available
Causal illusions occur when people perceive a causal relation between two events that are actually unrelated. One factor that has been shown to promote these mistaken beliefs is the outcome probability. Thus, people tend to overestimate the strength of a causal relation when the potential consequence (i.e. the outcome) occurs with a high probability (outcome-density bias). Given that children and adults differ in several important features involved in causal judgment, including prior knowledge and basic cognitive skills, developmental studies can be considered an outstanding approach to detect and further explore the psychological processes and mechanisms underlying this bias. However, the outcome density bias has been mainly explored in adulthood, and no previous evidence for this bias has been reported in children. Thus, the purpose of this study was to extend outcome-density bias research to childhood. In two experiments, children between 6 and 8 years old were exposed to two similar setups, both showing a non-contingent relation between the potential cause and the outcome. These two scenarios differed only in the probability of the outcome, which could either be high or low. Children judged the relation between the two events to be stronger in the high probability of the outcome setting, revealing that, like adults, they develop causal illusions when the outcome is frequent.
... In the contingency learning literature, learning is assessed by means of judgments collected at the end of the training session, just as we did in our study. Available evidence indicates that the wording of the question can affect substantially the judgment given by the participant [54,55]. In our case, we chose to formulate the question in terms of effectiveness (i.e., how effective the medicine was to heal the patients), instead of in terms of causality (i.e., how likely the medicine was the cause of the healings), because this wording seems easier to understand, more intuitive for participants, and more ecological, while retaining basically the same meaning. ...
Article
Full-text available
In the reasoning literature, paranormal beliefs have been proposed to be linked to two related phenomena: a biased perception of causality and a biased information-sampling strategy (believers tend to test fewer hypotheses and prefer confirmatory information). In parallel, recent contingency learning studies showed that, when two unrelated events coincide frequently, individuals interpret this ambiguous pattern as evidence of a causal relationship. Moreover, the latter studies indicate that sampling more cause-present cases than cause-absent cases strengthens the illusion. If paranormal believers actually exhibit a biased exposure to the available information, they should also show this bias in the contingency learning task: they would in fact expose themselves to more cause-present cases than cause-absent trials. Thus, by combining the two traditions, we predicted that believers in the paranormal would be more vulnerable to developing causal illusions in the laboratory than nonbelievers because there is a bias in the information they experience. In this study, we found that paranormal beliefs (measured using a questionnaire) correlated with causal illusions (assessed by using contingency judgments). As expected, this correlation was mediated entirely by the believers' tendency to expose themselves to more cause-present cases. The association between paranormal beliefs, biased exposure to information, and causal illusions was only observed for ambiguous materials (i.e., the noncontingent condition). In contrast, the participants' ability to detect causal relationships which did exist (i.e., the contingent condition) was unaffected by their susceptibility to believe in paranormal phenomena.
... In addition to active vs. passive roles of participants, there are many other variants that can be introduced in this task and that have been shown to affect the participants' estimations of causality. Examples include changing the wording of questions asked at the end of the experiment about the causal relationship (Crocker, 1982;Vadillo et al., 2005Vadillo et al., , 2011Collins and Shanks, 2006;De Houwer et al., 2007;Blanco et al., 2010;Shou and Smithson, 2015), the order in which the different trial types are presented (Langer and Roth, 1975;López et al., 1998), the frequency with which judgments are requested (Collins and Shanks, 2002;Matute et al., 2002), the description of the relevant events as causes, predictors, or effects (Waldmann and Holyoak, 1992;Cobos et al., 2002;Pineño et al., 2005), the temporal contiguity between the two events (e.g., Shanks et al., 1989;Wasserman, 1990;Lagnado and Sloman, 2006;Lagnado et al., 2007), and many other variables that fortunately are becoming well known. In the following sections, we will focus on the variables that seem to affect the illusion most critically in cases of null contingency. ...
Article
Full-text available
Illusions of causality occur when people develop the belief that there is a causal connection between two events that are actually unrelated. Such illusions have been proposed to underlie pseudoscience and superstitious thinking, sometimes leading to disastrous consequences in relation to critical life areas, such as health, finances, and wellbeing. Like optical illusions, they can occur for anyone under well-known conditions. Scientific thinking is the best possible safeguard against them, but it does not come intuitively and needs to be taught. Teaching how to think scientifically should benefit from better understanding of the illusion of causality. In this article, we review experiments that our group has conducted on the illusion of causality during the last 20 years. We discuss how research on the illusion of causality can contribute to the teaching of scientific thinking and how scientific thinking can reduce illusion.
... The only experiment that, as far as we know, has been conducted with humans using anticipatory behavior per se as the dependent variable was conducted by Blanco, Matute, and Vadillo (2010). They reported that anticipatory behavior in humans was driven by contingency. ...
Article
Full-text available
A stimulus is a reliable signal of an outcome when the probability that the outcome occurs in its presence is different from in its absence. Reliable signals of important outcomes are responsible for triggering critical anticipatory or prepara-tory behavior, which is any form of behavior that prepares the organism to receive a biologically significant event. Previous research has shown that humans and other animals prepare more for outcomes that occur in the presence of highly reliable (i.e., highly contingent) signals, that is, those for which that difference is larger. However, it seems reason-able to expect that, all other things being equal, the probability with which the outcome follows the signal should also affect preparatory behavior. In the present experiment with humans, we used two signals. They were differentially fol-lowed by the outcome, but they were equally (and relatively weakly) reliable. The dependent variable was preparatory behavior in a Martians video game. Participants prepared more for the outcome (a Martians' invasion) when the out-come was most probable. These results indicate that the probability of the outcome can bias preparatory behavior to occur with different intensities despite identical outcome signaling.
Article
Full-text available
Judgments of a treatment's effectiveness are usually biased by the probability with which the outcome (e.g., symptom relief) appears: even when the treatment is completely ineffective (i.e., there is a null contingency between cause and outcome), judgments tend to be higher when outcomes appear with high probability. In this research, we present ambiguous stimuli, expecting to find individual differences in the tendency to interpret them as outcomes. In Experiment 1, judgments of effectiveness of a completely ineffective treatment increased with the spontaneous tendency of participants to interpret ambiguous stimuli as outcome occurrences (i.e., healings). In Experiment 2, this interpretation bias was affected by the overall treatment-outcome contingency, suggesting that the tendency to interpret ambiguous stimuli as outcomes is learned and context-dependent. In conclusion, we show that, to understand how judgments of effectiveness are affected by outcome probability, we need to also take into account the variable tendency of people to interpret ambiguous information as outcome occurrences.
Thesis
Full-text available
Studies of people's beliefs about how much they control events have shown that people often overestimate the extent to which the result depends on their own behavior. Studies of people's beliefs about how much they control events have shown that people often overestimate the extent to which the result depends on their own behavior. The purpose of this study is to assess the relationship of emotional characteristics and formulation of the question on the illusion of control, depending on the desirable and undesirable results. In the study, it was assumed that the illusion of control depends on the amount of effort applied to achieve the result. It has also been suggested to reduce the illusion of control when asking a causal question in the case where the result is desirable and the participant acts to make that result appear, and in the case where the result is undesirable and the subject acts to prevent it from occurring. The influence of the cause-effect question and emotional characteristics on the value of the illusion of control, measured by the self-esteem of the subjects was not found. There was also no correlation between the amount of effort and the illusion of control.
Article
Most previous research on illusions of control focused on generative scenarios, in which participants' actions aim to produce a desired outcome. By contrast, the illusions that may appear in preventive scenarios, in which actions aim to prevent an undesired outcome before it occurs, are less known. In this experiment, we studied two variables that modulate generative illusions of control, the probability with which the action takes place, P(A), and the probability of the outcome, P(O), in two different scenarios: generative and preventive. We found that P(O) affects the illusion in symmetrical, opposite directions in each scenario, while P(A) is positively related to the magnitude of the illusion. Our conclusion is that, in what concerns the illusions of control, the occurrence of a desired outcome is equivalent to the nonoccurrence of an undesired outcome, which explains why the P(O) effect is reversed depending on the scenario.
Article
Full-text available
The research reported in this article replicated the well-established phenomenon of competition between causes (C) as well as the more controversial presence and absence of competition between effects (E). The test question was identified as a crucial factor leading to each outcome. Competition between causes was obtained when the test question asked about the probability of E given C, p(E/C), implicitly compared with the probability of E given some alternative cause, p(E/C'). competition between effects was obtained when the test question asked about p(C/E) implicitly compared with p(C/E'). Under these conditions, effects competed for diagnostic value just as causes competed for predictive value. Additionally, some conditions in which neither causes nor effects competed were identified. These results suggest a bidirectional and noncompetitive learning process, the contents of which can be used in different ways (competitively or noncompetitively and forward or backward) as a function of test demands.
Article
Full-text available
Conditioned suppression is a useful technique for assessing whether subjects have learned a CS-US association, but it is difficult to use in humans because of the need for an aversive US. The purpose of this research was to develop a non-aversive procedure that would produce suppression. Subjects learned to press the space bar of a computer as part of a video game, but they had to stop pressing whenever a visual US appeared, or they would lose points. In Experiment 1, we used an A+/B- discrimination design: The US always followed Stimulus A and never followed Stimulus B. Although no information about the existence of CSs was given to the subjects, suppression ratio results showed a discrimination learning curve-that is, subjects learned to suppress responding in anticipation of the US when Stimulus A was present but not during the presentations of Stimulus B. Experiment 2 explored the potential of this preparation by using two different instruction sets and assessing post-experimental judgements of CS A and CS B in addition to suppression ratios. The results of these experiments suggest that conditioned suppression can be reliably and conveniently used in the human laboratory, providing a bridge between experiments on animal conditioning and experiments on human judgements of causality.
Article
Associative and statistical theories of causal and predictive learning make opposite predictions for situations in which the most recent information contradicts the information provided by older trials (e.g., acquisition followed by extinction). Associative theories predict that people will rely on the most recent information to best adapt their behavior to the changing environment. Statistical theories predict that people will integrate what they have learned in the two phases. The results of this study showed one or the other effect as a function of response mode (trial by trial vs. global), type of question (contiguity, causality, or predictiveness), and postacquisition instructions. That is, participants are able to give either an integrative judgment, or a judgment that relies on recent information as a function of test demands. The authors concluded that any model must allow for flexible use of information once it has been acquired.
Article
Recently, the study of biological function has been reaccepted as a legitimate focus of research in the field of animal learning. This "new" functionalism suffuses 2 distinct perspectives with which researchers approach the study of Pavlovian conditioning. Those who adopt the ecological perspective explore the role of conditioning within functional, naturally occurring categories of behavior, for example, intraspecific defense, reproduction, or food recognition. From this perspective, the central question is, In what ways does conditioning contribute to animals' ability to survive and reproduce? For those researchers who explore the cause rather than the function of conditioning, the central question is, How, and under what circumstances, does conditioning occur? Although, historically, those who studied causal mechanisms eschewed functional considerations, close examination of the new cognitive perspective reveals decidedly functional themes. A brief review of research originating in each of these 2 perspectives demonstrates the ways in which they increasingly are finding common ground in a "new" study of function.
Article
MartiansV2 is both a language syntax in which experiments can be written and an implementation of this syntax in a runtime application that, when fed a valid experiment text file, will execute the given experiment. It is based on the original Martians preparation, which has proven a valuable tool for assessing human online-conditioned suppression performance through research on a wide array of learning phenomena. This article can be read as a manual, both for using the Martians paradigm in general and for getting started with MartiansV2.
Article
The term pseudocontingency (PC) denotes the logically unwarranted inference of a contingency between 2 variables X and Y from information other than pairs of xi, yi observations, namely, the variables' univariate base rates as assessed in 1 or more ecological contexts. The authors summarize recent experimental evidence showing that PCs can play a pivotal role in many areas of judgment and decision making. They argue that the exploitation of the informational value of base rates underlying PCs offers an alternative perspective on many phenomena in the realm of adaptive cognition that have been studied in isolation so far. Although PCs can lead to serious biases under some conditions, they afford an efficient strategy for inductive inference making in probabilistic environments that render base-rate information, rather than genuine covariation information, readily available.
Article
2 EXPERIMENTS WITH MALE SPRAGUE-DAWLEY RATS INDICATE THAT CS-UCS CONTINGENCY IS AN IMPORTANT DETERMINANT OF FEAR CONDITIONING AND THAT PRESENTATION OF UCS IN THE ABSENCE OF CS INTERFERES WITH FEAR CONDITIONING. IN EXP. I, EQUAL PROBABILITY OF A SHOCK UCS IN THE PRESENCE AND ABSENCE OF A TONE CS PRODUCED NO CONDITIONED EMOTIONAL RESPONSE SUPPRESSION TO CS; THE SAME PROBABILITY OF UCS GIVEN ONLY DURING CS PRODUCED SUBSTANTIAL CONDITIONING. IN EXP. II, WHICH EXPLORED 4 DIFFERENT PROBABILITIES OF UCS IN THE PRESENCE AND ABSENCE OF CS, AMOUNT OF CONDITIONING WAS HIGHER THE GREATER THE PROBABILITY OF UCS DURING CS AND WAS LOWER THE GREATER THE PROBABILITY OF UCS IN THE ABSENCE OF CS; WHEN THE 2 PROBABILITIES WERE EQUAL, NO CONDITIONING RESULTED.
Article
Recently, the study of biological function has been reaccepted as a legitimate focus of research in the field of animal learning. This "new" functionalism suffuses 2 distinct perspectives with which researchers approach the study of Pavlovian conditioning. Those who adopt the ecological perspective explore the role of conditioning within functional, naturally occurring categories of behavior, for example, intraspecific defense, reproduction, or food recognition. From this perspective, the central question is, In what ways does conditioning contribute to animals' ability to survive and reproduce? For those researchers who explore the cause rather than the function of conditioning, the central question is, How, and under what circumstances, does conditioning occur? Although, historically, those who studied causal mechanisms eschewed functional considerations, close examination of the new cognitive perspective reveals decidedly functional themes. A brief review of research originating in each of these 2 perspectives demonstrates the ways in which they increasingly are finding common ground in a "new" study of function.