ArticlePDF Available

Abstract and Figures

Several classic studies have concluded that the accuracy of identifying uncontrollable situations depends heavily on depressive mood. Nondepressed participants tend to exhibit an optimistic illusion of control, whereas depressed participants tend to better detect a lack of control. Recently, we suggested that the different activity levels (measured as the probability of responding during a contingency learning task) exhibited by depressed and nondepressed individuals is partly responsible for this effect. The two studies presented in this paper provide further support for this mediational hypothesis, in which mood is the distal cause of the illusion of control operating through activity level, the proximal cause. In Study 1, the probability of responding, P(R), was found to be a mediator variable between the depressive symptoms and the judgments of control. In Study 2, we intervened directly on the mediator variable: The P(R) for both depressed and nondepressed participants was manipulated through instructions. Our results confirm that P(R) manipulation produced differences in the participants' perceptions of uncontrollability. Importantly, the intervention on the mediator variable cancelled the effect of the distal cause; the participants' judgments of control were no longer mood dependent when the P(R) was manipulated. This result supports the hypothesis that the so-called depressive realism effect is actually mediated by the probability of responding.
Content may be subject to copyright.
Mediating Role of Activity Level in the Depressive
Realism Effect
Fernando Blanco*, Helena Matute, Miguel A. Vadillo
Departamento de Fundamentos y Me
´todos de la Psicologı
´a, University of Deusto, Bilbao, Spain
Several classic studies have concluded that the accuracy of identifying uncontrollable situations depends heavily on
depressive mood. Nondepressed participants tend to exhibit an optimistic illusion of control, whereas depressed
participants tend to better detect a lack of control. Recently, we suggested that the different activity levels (measured as the
probability of responding during a contingency learning task) exhibited by depressed and nondepressed individuals is
partly responsible for this effect. The two studies presented in this paper provide further support for this mediational
hypothesis, in which mood is the distal cause of the illusion of control operating through activity level, the proximal cause.
In Study 1, the probability of responding, P(R), was found to be a mediator variable between the depressive symptoms and
the judgments of control. In Study 2, we intervened directly on the mediator variable: The P(R) for both depressed and
nondepressed participants was manipulated through instructions. Our results confirm that P(R) manipulation produced
differences in the participants’ perceptions of uncontrollability. Importantly, the intervention on the mediator variable
cancelled the effect of the distal cause; the participants’ judgments of control were no longer mood dependent when the
P(R) was manipulated. This result supports the hypothesis that the so-called depressive realism effect is actually mediated
by the probability of responding.
Citation: Blanco F, Matute H, Vadillo MA (2012) Mediating Role of Activity Level in the Depressive Realism Effect. PLoS ONE 7(9): e46203. doi:10.1371/
Editor: Lin Lu, Peking University, China
Received May 28, 2012; Accepted August 28, 2012; Published September 27, 2012
Copyright: ß2012 Blanco et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Support for this research was provided by the Direccio
´n General de Investigacio
´n of the Spanish Government (Grant PSI2011-26965) and the
Departamento de Educacio
´n, Universidades e Investigacio
´n of the Basque Government (Grant PI2012-56). The funders had no role in study design, data collection
and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail:
Far from being optimal reasoners, people are known to exhibit
certain cognitive biases or systematic errors when performing
certain tasks. One of the most extensively studied cognitive biases
is the illusion of control, which consists of a belief that one is able
to exert control over outcomes that are actually uncontrollable.
This effect occurs mainly when the outcomes are of appetitive
valence and occur frequently [1–3].
A very influential paper by Alloy and Abramson [1] showed that
the illusion of control was stronger in nondepressed participants.
In that study, both nondepressed and mildly depressed students
were allowed to decide during each trial whether to press a button
to turn on a light bulb. The light bulb’s onset was actually
independent of the pressing of the button (i.e., it was uncontrol-
lable). During a training phase, the participants were able to press
the button as often as they wanted, and the light came on during
some of the trials but not others. Afterward, the participants were
asked about the degree of control they had exerted over the light
onset; that is, they were asked about the perceived contingency
between their responses and the outcomes. Interestingly, Alloy and
Abramson [1] found that mildly depressed participants were better
at detecting the absence of control over the light than
nondepressed participants were; nondepressed participants sys-
tematically overestimated the contingency between their actions
and the uncontrollable outcome, thereby exhibiting an illusion of
control [4–6]. Alloy and Abramson’s finding soon became known
as ‘‘depressive realism’’ [7] and has been replicated several times
using different tasks and procedures [8–12].
A variety of hypotheses have been proposed to account for this
puzzling finding (see a review by Ackerman & DeRubeis [13], and
recent proposals by different authors [9–11,14]). An examination
of the published literature leads to the impression that the
presumed realism of depressed people may be a complex,
multicausal phenomenon that involves many variables and
demands further research.
One variable that has been shown to be a source of bias in
contingency judgments is the activity level of the participants, as
determined by the response frequency [6,15,16]. Participants who
take an active approach to this type of experiment become
exposed to frequent, yet accidental, pairings between their
responses and the uncontrollable outcome. These co-occurrences
lead them to believe that they are indeed causing the outcome; as a
result, they develop an illusion of control.
Inspired by this rationale, a recent study by Blanco et al. [9]
suggested that the passivity associated with depressive symptoms
does, to some extent, contribute to realism in detecting the
uncontrollability in the standard contingency-learning tasks
normally used in studies of depressive realism [1]. According to
Blanco et al. [9], nondepressed participants tend to spontaneously
respond more frequently during the training phase. This means
that in a typical contingency learning task, nondepressed
participants decide to press the button on more trials than do
depressed participants, who very often take a more passive
PLOS ONE | 1 September 2012 | Volume 7 | Issue 9 | e46203
approach to the task [17]. This passive strategy leads to few
accidental co-occurrences between the response and the outcome
and thus might prevent the development of an illusion of control.
In line with these predictions, Blanco et al. showed that mildly
depressed participants were both more passive and more realistic
than nondepressed participants in an experimental task involving
uncontrollable outcomes, similar to that of Alloy and Abramson
[1]. In addition, the authors also reported that passivity was
correlated with accuracy in detecting the absence of control. The
latter result converges with Matute’s [6] findings that the more
active participants develop stronger illusions of control, although
Matute did not include a measure of depressive symptoms in her
study. As a conclusion, Blanco et al. [9] suggested that the
depressive realism effect could partially be due to the passivity that
depressive participants exhibit in a variety of experimental settings.
Thus, depressive symptoms would be the distal cause of the
realistic detection of uncontrollability via a more proximal cause,
depressive passivity, defined as a lower probability of responding
during a contingency learning task.
Nonetheless, the claim that passivity associated with depressive
symptoms plays a relevant causal role in the depressive realism
effect needs to be supported by careful data analysis and, more
importantly, by evidence from experimental designs in which the
appropriate manipulations are carried out. Thus, our current aim
is to complement the work started by Blanco et al. [9] by using two
strategies, statistical and experimental. First, we conducted an
exact replication of Blanco et al.’s study [9], but we included a
proper mediational analysis. This statistical technique is intended
to test the hypothesis by isolating the direct effect of depressive
symptoms on control judgments while partialling out the effect of
the probability of responding, P(R). In the second study, we
conducted an experimental manipulation to further test the
mediational hypothesis. Specifically, we intervened directly on
the mediator variable, P(R). Because the hypothesized proximal
cause of the illusion, P(R), was manipulated externally, if the
mediational hypothesis were true, we would observe that the
hypothesized distal cause (i.e., depressive symptoms) and the
outcome variable (i.e., control judgments) become mutually
independent. In other words, we expect that the depressive
realism effect should disappear in our experimental study when we
directly intervene on the mediator between depressive symptoms
and control judgments. The use of this intervention strategy to
reveal causal links is described in the causal reasoning literature
[18,19] as based on the Markov condition for Bayesian networks,
which states that a variable is probabilistically independent of any
other variable (except for its descendants) conditional on its
parents. In our case, if the mediational model is true, intervening
on the parent of the outcome variable (i.e., its proximal cause) will
render the outcome variable and the distal cause independent
from each other.
Study 1
Study 1 was an exact replication of Blanco et al. [9]. Therefore,
we expect that participants who are more depressed would show
more passive behaviors (i.e., fewer responses) during the training
phase and would exhibit more accurate and realistic control
judgments. This prediction stems from previous findings and from
our interpretation of the depressive realism effect as a consequence
of the lower probability of responding. Importantly, Study 1
features a more appropriate statistical procedure than the one that
Blanco et al. [9] used. Specifically, in the current study we make
use of a mediation analysis aimed at testing whether the
probability of responding is a significant mediator between
depressive symptoms and control judgments.
Participants and apparatus. Fifty Psychology students at
the University of Deusto agreed to participate in the study as an
optional and voluntary activity in their course. The students were
not asked to give their name or any other personal information,
and were given neither course credit nor monetary compensation
for this activity. Data from one participant were excluded because
he/she did not respond on any trial, thus making it impossible to
judge whether he/she was paying attention. The study took place
during a collective session in a large computer room, and was
carried out by a teaching assistant different from the course
The study used the Flashes task, programmed in JavaScript.
Previous studies have demonstrated that this task is particularly
suitable for assessing the illusion of control [20] and the depressive
realism effect [9] both on the Internet and in the laboratory.
Ethics statement. In Study 1 and Study 2, the behavioral
and verbal data generated by the participants were sent
anonymously to the experimenters’ e-mail address through the
Internet browser. In agreement with the ethical standards for
human research on the Internet [21,22], the participants were
informed before the session that in no case they would be asked to
provide any personal information (e.g., name, phone, address, e-
mail), that their data would not be identifiable, and that they
would be allowed to terminate the study by closing the Internet
browser window at any moment without penalty, if they wished so.
In addition, right after the study finished, a screen requested the
participants to give permission to use the data they had just
generated. The participants agreed to grant this permission by
clicking a button labeled ‘‘Send data’’, which immediately
submitted the data anonymously to the experimenters’ database.
Those participants not willing to send the information after the
experiment were instructed to click a button labeled ‘‘Finish’’,
which immediately deleted the data. Finally, we did not use
cookies, keyloggers or any other software to covertly obtain
information from the participants.
The funding agency for this research (i.e., Spanish Ministry of
Science and Innovation) evaluated the scientific and practical
implications of the procedure used in these studies prior to the
approval of the research grant, and did not request any extra
formal evaluation from an ethical review board (neither before the
allocation of funds nor in subsequent renewals). This was in
agreement with the code of ethics of the Spanish Psychological
Association [23] (Chapter IV, Articles 33 to 38), which does not
state the compulsory nature of formal ethical approval before
carrying out this type of innocuous psychological research. In our
procedure, the data that the participants provided were anony-
mous and unidentifiable, the stimuli and materials were harmless
and emotionally neutral, the goal of the study was transparent, and
the task involved no deception.
Procedure. The procedure was identical to the one used by
Blanco et al. [9]. A computerized version of Beck’s Depression
Inventory (BDI [24]; Spanish adaptation by Conde & Useros [25])
was administered prior to the study, in order to assess the
participants’ depressive symptoms. Rather than establishing an
arbitrary cutoff point to split the sample into mildly depressed and
nondepressed participants, as has usually been done in previous
studies [1,9–11], the direct BDI score was taken as a continuous
variable. Acknowledging the actual nature of the variable, which is
indeed continuous, permits a less biased and more powerful
approach to data analysis (for a more elaborated argument on this
Activity Level and Depressive Realism
PLOS ONE | 2 September 2012 | Volume 7 | Issue 9 | e46203
methodological issue, see Cohen [26]). Using BDI scores as a
continuous dimension, rather than as a categorical one, raises a
terminology-related issue that must be addressed. Given that our
sample is not clinical, the term ‘‘depressed’’ seems inadequate.
However, once we acknowledge that our measure of depressive
symptoms is continuous in nature, it seems reasonable to use the
term ‘‘depressed’’ not as a clinical label, but as an indicator of a
given participant’s or group’s higher BDI scores relative to others
in the sample. Therefore, one may say that one participant is
‘‘more depressed’’ than other, while none of them is clinically
depressed. Thus, through the manuscript, we use the terms
‘‘depressed’’ and ‘‘nondepressed’’ not as indicators of a clinical
diagnosis, but as a indicators of relative individual or group
differences in the measure of the depressive symptoms.
After recording the BDI score, the program displayed written
instructions for the task. The participants were told that a flash
would appear on the screen from time to time and last for 1
second. Then, the screen would fade to black. As the instructions
described, the participants’ goal was to make the flash appear
again whenever it stopped by pressing the space bar on the
keyboard. That is, the termination of the flash signaled the
opportunity to respond to make it appear again. After the
instructions, a series of fifty flashes with a fixed duration of 1
second was presented, interspersed with fifty black-screen inter-
vals. The lengths of these black intervals were programmed to be
either short (1 second) or long (5 seconds) according to a
prescheduled randomized sequence. This means that, on a given
trial during which the participant had responded (by pressing the
space bar) right after the previous flash stopped, the flash could
follow either immediately (i.e., with a short, 1-second interval),
thus leading to an accidental concurrence between the response
and the flash, or after a delay (i.e., a long, 5-second interval). The
instructions specified that the flashes would appear after a short
interval of 1 second whenever the participant succeeded at
producing them. They also specified that the flashes would appear
after a delay of 5 seconds whenever the participant failed. In other
words, the trials in which the flash appears after a short 1-second
interval are given reinforcing value through instructions, whereas
the trials in which the flash appears after a long delay (5 seconds)
are intended to serve as unreinforced trials.
Given that the sequence of flashes was prescheduled and
therefore response-independent, the onset of the flashes was
actually uncontrollable for the participants. Nevertheless, acciden-
tal concurrences between response and flash were expected. These
concurrences, if frequent, could lead to the impression that flashes
were under the participants’ control. The fact that 38 out of 50
trials (76%) included a short black interval and only 12 trials (24%)
included a long black interval (i.e., 76% of trials were potentially
reinforced) was aimed at favoring the illusion of control, based on
previous studies showing that a high probability of reinforcement
produces stronger illusions [1,2]. The predefined sequence of trials
was identical to that used by Blanco et al. [9] to render a fair
comparison of the two studies.
The activity level of every participant was measured by
computing the probability of responding, P(R), during the training
phase. This value results from dividing the number of trials in
which a response (a key press) was detected by the total number of
trials (50), hence the variable ranges between 0 and 1.
Finally, the participants’ perceived control over the flashes was
assessed via a control judgment after the training phase. The test
question, translated from the original in Spanish, read as follows:
‘‘To what extent do you think that the onset of the flashes
depended on what you did?’’ The answers were given by clicking
on a scale ranging from 0 (‘‘Absolutely not’’) to 100 (‘‘Absolutely’’).
Given that the flashes were presented in a preprogrammed order,
one can assume that any judgment higher than zero was biased up
to a certain extent.
Results and Discussion
Table 1 contains the descriptive statistics (mean, standard error
of the mean, range and median) for the three variables assessed:
direct BDI score, P(R), and judgment of control. Instead of
dichotomizing the continuous variables, as is usually done in this
literature, we tested a potential mediational structure by means of
the method described by Baron and Kenny [27]. In the proposed
mediational structure, the BDI scores would affect the control
judgments (i.e., produce the depressive realism effect) via the
intermediate variable, as shown in Figure 1B. Following Baron
and Kenny’s procedure, we first assessed the direct effect of the
model, proving that BDI scores marginally predicted the
judgments of control, b=2.27, t(48) = 1.95, p= .057 (Path c in
Figure 1A). The negative coefficient implies that the higher the
participant’s BDI score, the lower (and hence, more realistic) his or
her control judgment was (see Figure 2). Therefore, we replicated
the depressive realism effect described by Alloy & Abramson [1].
Next, we found that BDI scores also predicted the P(R), b=2.52,
t(48) = 4.26, p,.001 (Path a in Figure 1B). The participants who
were more depressed also responded less frequently during the
task, in line with the findings of previous studies [9]. In the final
step, we conducted a multiple regression analysis with BDI scores
and P(R) as predictors and judgments as the predicted variable.
This revealed that P(R) was a reliable predictor of the judgments
even when controlling for the BDI scores, b= .40, t(47) = 2.57,
p,.05 (Path b in Figure 1B), whereas the BDI scores failed to
predict these judgments, b=2.06, t(47) = 0.40, p= .68 (the direct
effect depicted as Path c’ in Figure 1B). The Sobel test [28]
confirmed that the variance in the judgments that the BDI scores
were able to predict (i.e., the total effect) was in fact accounted for
by the indirect path via a mediator variable, P(R), z= 2.22, p,.05.
Once the effect that the proximal cause, P(R), had on judgments
was partialed out, there was little variance left in the judgments
that could be predicted directly by the distal cause (namely, the
BDI scores). This finding is consistent with our initial mediational
Table 1. Descriptive statistics.
BDI scores Judgments P(R)
Study 1 (N= 49) Mean 6.92 37.45 .75
SEM 0.88 3.82 .03
Range 0–25 0–85 .12–1
Median 6 45 .88
Study 2: Analytic
instructions (N = 95)
Mean 9.33 26.05 .48
SEM 0.82 3.24 .03
Range 0–36 0–100 .02–1
Median 7 15 .48
Study 2: Naturalistic
instructions (N = 103)
Mean 9.60 30.53 .65
SEM 0.84 2.91 .02
Range 0–40 0–100 .04–1
Median 7 20 .70
Activity Level and Depressive Realism
PLOS ONE | 3 September 2012 | Volume 7 | Issue 9 | e46203
Study 2
Blanco et al. [9] and (more clearly) its exact replication in Study
1 of the current report both suggested that a causal chain model
links depressive symptoms (as assessed by the BDI) to judgments of
control via a mediator variable (the activity level), as shown in
Figure 1B. To further investigate the mediation of the P(R)
between the BDI scores and the judgments, we conducted Study 2.
Here, we intervened by manipulating the P(R) via instructions,
independent of depressive symptoms (as shown in Figure 3). A
direct manipulation of the P(R) breaks the causal chain because it
prevents mood from affecting P(R). Consequently, in Study 2, we
should find that the P(R), and not the distal cause (BDI scores), is
the variable that best predicts judgments of control.
This strategy aims to reveal causal links by means of
interventions, and is grounded in the causal reasoning literature
[18,19], which typically embeds the variables into causal networks
similar to those depicted in Figures 1 and 3. Under the
mediational hypothesis, when an intervention is performed on a
mediator variable the effect of the distal cause (called the ‘‘parent
variable’’ in the causal networks literature) on subsequent elements
(the ‘‘descendant variables’’) of the chain is cancelled due to the
Markov condition. Therefore, this manipulation serves as a test of
the mediational hypothesis of depressive realism.
In addition, the choice of P(R) as the independent variable in
Study 2 allows for disambiguation between our mediational
hypothesis and an alternative causal model that may also be at the
basis of the depressive realism effect. According to a common-
cause model, both the P(R) and the control judgments would be
directly affected by depressive symptoms. Crucially, the manipu-
lation of the P(R) under such an alternative hypothesis would not
eliminate the depressive realism effect because the intervention
would be performed on a terminal node (i.e., a variable without
descendants), and the link between BDI scores and judgments
Figure 1. Mediational model tested in Study 1. The letters a, b, c,
and c’ depict the paths between the three variables, which are
weighted by the standardized regression coefficients. One or two
asterisks indicate a significant coefficient (p,.05 or p,.01, respectively),
while n.s. means that the coefficient failed to reach the significance
level (p..05). Panel (A): Path c corresponds to the predictive link
between BDI scores and the judgments (i.e., the total effect in the
model). The negative coefficient indicates that we replicated the
depressive realism effect. Panel (B): The total effect (Path c above) can
be decomposed into two components, the direct effect and the indirect
effect. Paths a and b represent the indirect effect that operates via the
mediator, P(R). Path c’ represents the direct effect of BDI scores on the
judgments, that is, the amount of predictive power left after the
mediational effect of P(R) has been partialed out. Overall, the
mediational analyses support a causal chain model in which BDI scores
produce depressive realism indirectly via the mediator variable, P(R).
Figure 2. Scatterplot depicting the judgments as a function of BDI scores in Study 1. The scatterplot shows the judgments of control
(vertical axis) as a function of the direct BDI scores (horizontal axis). The line fitting the data points displays a negative slope, indicating that the
participants with higher BDI scores were more accurate in their judgments, thus replicating the depressive realism effect.
Activity Level and Depressive Realism
PLOS ONE | 4 September 2012 | Volume 7 | Issue 9 | e46203
would remain statistically intact. That is, concerning the manip-
ulation of P(R), the predictions of the two alternative models differ
Participants and apparatus. A sample of 209 anonymous
Internet users volunteered to participate in the study through our
laboratory’s webpage ( The experimental
task was identical to that used in Study 1 and in previous
publications [9,20]. Previous studies conducted in our own
laboratory and elsewhere showed that samples collected on the
Internet are as reliable as those collected in the laboratory,
specifically college students [20]. A frequently raised concern
about Internet-based studies is the possibility of multiple
submission of the same participant. We acknowledge that it is
indeed impossible to assure that no participant in our sample took
part twice in Study 2. However, according to most studies on this
issue [29,30], multiple submission is a rather uncommon event (of
virtually negligible impact on the sample of 209 Internet users in
Study 2) [29,30]. Birnbaum [29] concluded that multiple
submission ‘‘is not a real problem’’ in Internet-based research.
Note also that, as mentioned above, we tested the reliability of this
particular procedure (i.e., the flashes task) in previous studies in
which we compared a sample of Internet users (collected in similar
conditions to Study 2) with another sample comprising college
students, and found the same results in both of them [20].
Moreover, convergent results concerning the depressive realism
effect were also found in Blanco et al. [9], with a sample of
Internet users, and in Study 1, which was conducted with college
students, despite the fact that multiple submission was only
possible through the Internet.
The participants were randomly assigned to one of two groups,
which differed only in the instructions they received before the
training phase. Eleven participants were excluded from the
analyses because they did not respond in even a single trial during
the experiment, which made the data they provided unreliable
(there is no way to determine whether they were paying attention
to the experiment). Therefore, the final sample was reduced to 198
volunteers, 103 in the Naturalistic Instructions group and 95 in the
Analytic Instructions group.
Procedure. The procedure was identical to that used in
Study 1 except for the instructions given before the training. The
participants were presented with either of two instructional sets,
the Naturalistic instructions or the Analytic instructions. The
participants in the Naturalistic instructions group were encouraged
to press the space bar frequently to make the flashes of light appear
on the screen, whereas the participants in the Analytic instructions
group were told to press the space bar in only 50% of the trials to
determine whether they could control the flashes with their
responses. The two instructional sets were based on the
instructions from Matute [6], in which the same manipulation of
the P(R) was conducted successfully.
Results and Discussion
The descriptive statistics for BDI, P(R), and judgments are
summarized in Table 1. To begin, we ensured that there were no
significant differences in the BDI scores between the two groups (F
,1), thus avoiding a potential source of sample bias. As planned,
the instructional sets produced higher P(R) levels in the
Naturalistic group than in the Analytic group, F(1, 196) = 21.15,
p,.001. This manipulation was conducted independent of the
participants’ depressive symptoms. As expected, the BDI scores
were unable to affect the P(R) in Study 2, b=2.03, t(196) = 0.48,
p= .63. The manipulation of the activity level successfully
prevented the depressive passivity that we found in Study 1. In
Figure 3, this is displayed as a crossed-out path between the BDI
scores and the P(R). As we predicted from the mediational
hypothesis, two linear regression analyses showed that BDI scores
failed to predict the judgments, both when the two groups were
collapsed, b= .01, t(196) = 0.19, p= .85, and when the instructions
were controlled as a factor in the model, b= .01, t(195) = 0.17,
p= .86 (observe the flat slopes in the scatterplot of Figure 4). Thus,
intervening on the mediator variable P(R) resulted in the
abolishment of the depressive realism effect.
In contrast, the P(R) was a good predictor of the judgments,
b= .33, t(196) = 4.86, p,.001, even when we controlled for the
instructions by entering them in the model, b= .34, t(195) = 4.75,
p,.001. Moreover, once the P(R) effect was partialed out, the
direct effect of the instructions on judgments was not significant,
b=2.03, t(195) = 0.46, p= .64. This latter result indicates that the
variance in the judgments was due to the P(R) rather than any
potential direct influence of the instructions (e.g., different
expectations directly induced by the different instruction versions).
To sum up, Study 29s results converge with the conclusions from
Study 1 by highlighting that P(R) predicts the variance in the
judgments to a greater extent than the BDI scores do. This is
consistent with the idea that P(R) is the proximal factor that
mediates between depressed symptoms and judgments of control.
Consistent with this mediational hypothesis, when we intervened
directly on the P(R), the mood measure was unable to account for
any variance in the judgments. Furthermore, the latter finding is
not predicted by the alternative hypothesis in which both P(R) and
judgments are directly dependent on the depressive symptoms.
General Discussion
The intriguing finding by Alloy and Abramson [1] opened a
fertile research field that soon spread from basic learning literature
to social and applied psychology (see reviews [13,14,31,32] and a
Figure 3. Diagram of the intervention in Study 2. The paths
represent the links between the variables of interest, weighted by the
provided standardized regression coefficients. One or two asterisks
indicate a significant coefficient (p,.05 or p,.01, respectively), while
n.s. means that the coefficient failed to reach the significance level
(p..05). In addition, crossed-out paths indicate that the two connected
variables are not correlated to each other. First, the diagram shows that
the direct intervention on the P(R) cancels out the otherwise natural
effect of mood on activity level (i.e., the depressive passivity reported in
Study 1 and in Blanco et al., 2009), as expected. Consequently, and
given the mediational structure that we described in Study 1 (see
Figure 1B), judgments are no longer predicted by BDI scores (i.e., the
intervention renders the BDI scores and the judgments mutually
independent, conditional on the mediator). This supports the media-
tional hypothesis in which the effect of mood on judgments is mainly
due to the mediation of P(R).
Activity Level and Depressive Realism
PLOS ONE | 5 September 2012 | Volume 7 | Issue 9 | e46203
recent meta-analysis [33]). According to the numerous works
published on the topic, many different variables may modulate the
depressive realism effect, which explains why cannot always be
replicated. Some of the previously studied variables include the
valence of the outcomes [1,34], the presence of an observer [35],
the outcome frequency [1], and the length of the intertrial interval
[10,11]. Recently, Blanco et al. [9] suggested that yet another
factor contributes to the so-called depressive realism effect. As they
pointed out, depressed participants’ lack of responding during the
training phase prevents the illusion of control by reducing the
frequency of accidental concurrences between the participant’s
responses and the uncontrollable outcome. In contrast, the
nondepressed participants (who are usually more active and
respond frequently during the training phase) are prone to develop
an illusion of control via the formation of a spurious association
between the two events (see Blanco et al. [15] and Matute [6] for
demonstrations of the effect of the response frequency on the
illusions of control). Not only is the P(R) effect a robust empirical
finding, it is also consistent with the predictions of leading
associative learning theories, e.g., the Rescorla-Wagner model
[36]; see Matute, Vadillo, Blanco, and Musca [37] for computer
simulations illustrating this point).
Blanco et al. [9] showed that, in an uncontrollable task,
depressed participants were both more passive and more accurate
than nondepressed participants, while the response frequency
correlated positively with the illusion of control. In Study 1, we
conducted an exact replication of Blanco et al. [9] featuring a
mediational analysis aimed at revealing the causal structure
connecting the three variables of interest. This analysis confirmed
that higher BDI scores correlated with more accurate judgments of
control (i.e., depressive realism) and that probability of responding,
P(R), mediated the effect between depressive symptoms and
judgments. The resulting causal model can be depicted as a chain,
with BDI scores as the distal cause and the P(R) as a more
proximal cause of the variance in the judgments of control
(Figure 1B). Crucially, depressive symptoms affected the judg-
ments only via the mediator variable, P(R). In addition, while the
study conducted by Blanco et al. [9] used a sample of anonymous
Internet users, Study 1 in the current report was conducted with
college students, and both yielded similar results (the former
population could be assumed to be relatively heterogeneous in age,
sex, etc.). In summary, the results from Study 1 suggest that the
findings of Blanco et al. [9] are robust and partially generalisable
across populations, while complementing them with additional
statistical evidence of the mediator role of P(R) in the depressive
realism effect.
Study 2 provides converging evidence for the latter finding. The
same causal structure was tested, but through a different strategy
(experimental intervention). Study 2 used exactly the same tasks
and parameters that were used in Blanco et al.’s work [9] and
Study 1, except for the instructions that the participants received
before the training phase. These instructions were intended to
manipulate the probability of responding, in a manner similar to
that employed in previous studies [6]. As expected, the group that
received the Analytic instructions (which included the requirement
to respond on 50% of the trials) responded less frequently than the
group that was presented with the Naturalistic instructions, which
included encouragement to respond frequently.
Aligning with Blanco et al. [9], the only predictor of judgments
was the probability of responding. Thus, when we prevented
depressive symptoms from influencing the level of responding
(because participants were explicitly instructed to respond with
Figure 4. Scatterplot depicting judgments as a function of BDI scores in Study 2. Three regression lines are plotted: The shaded line shows
the overall relation between BDI scores and judgments; the full black line and the dashed black line refer to the Analytic and Naturalistic groups,
respectively. Note the flatness of the three slopes (in fact, no coefficient was significantly different from zero; see main text).
Activity Level and Depressive Realism
PLOS ONE | 6 September 2012 | Volume 7 | Issue 9 | e46203
either high or low frequency), it was the response frequency and
not the depressive symptoms that predicted the judgments: A
higher probability of responding led to a stronger illusion of
control, whereas a lower probability of responding led to better
accuracy at detecting the uncontrollability of the outcome. We
highlight that, apart from the instructional manipulation, the task
was identical to the one in which the depressive realism effect was
shown in Blanco et al. [9] and in Study 1. In addition, the sample
size of Study 2 (N= 198) was much larger than that of Study 1
(N= 49) and Blanco et al. (N= 66), hence minimizing the
possibility that the absence of an effect of depressive symptoms
was due to a lack of statistical power. Then, we conclude that
depressive realism vanished when depressive symptoms were not
allowed to influence the response frequency because the latter was
directly manipulated.
It is important to note that our account that the depressive
realism effect is mediated by the probability of responding is not
incompatible with other proposed mechanisms based on cognitive
or behavioral factors. Thus, it is not intended to rule out the
complex and multicausal view on the phenomenon that some
experts support [13]. The clear and distinctive prediction that can
be derived from our proposal is that any manipulation aimed at
reducing the frequency of responses should also lead to the more
accurate detection of uncontrollability, be it depressive symptoms
(as in Study 1 and Blanco et al. [9]), instructions (as in Study 2),
fatigue, motivational factors, an analytical rather than naturalistic
approach to the problem, or other means.
Moreover, our mediational hypothesis allows a theoretical
account of depressive realism in terms of contingency learning. It is
important to mention that depressive realism has been recently
studied in this domain [10,11,14]. Most theoretical models of
learning, such as the Rescorla-Wagner model [36], do not include
any explicit statement about the effects of motivational or mood-
related factors on the learning of associations between stimuli. If
these factors are reconceptualized, however, as distal causes
mediated by the P(R), as we argue here, then some contingency
learning models may indeed account for depressive realism and
related phenomena because the effect of the mediator variable,
P(R), can be easily addressed by these theories. For instance, some
contingency learning models would claim that the illusion of
control displayed by nondepressed/active participants in our
studies could be explained as the result of granting uneven
importance to different types of trials during the training phase
(i.e., cell weighting). That is, when evaluating contingency, the
participants give different weights to each type of events they are
exposed to, and usually consider concurrences of the desired
outcomes with their responses to be the most important event [38–
40]. In more general domains (e.g., social psychology), people are
also known to base their beliefs and decisions on confirmatory
evidence, neglecting disconfirmatory information [41,42] in a way
similar to the one we are describing here. This idea is also implicit
in the Rescorla-Wagner model [36], given the widely accepted
assumption that the learning rate parameter possesses a greater
value in response-present trials than in response-absent trials. As a
result of this asymmetry, the associative strength of the response
increases more rapidly because of the accidental concurrences
than it decreases because of the response-no outcome trials.
Furthermore, note that for those participants responding with high
probability, the accidental response-outcome concurrences are
actually the most frequent type of trial [37].
To sum up, the predictions of these two variants of the
contingency learning theory (cell weighting-based and associative)
converge when they are applied to our two current studies: They
both predict that increasing the response frequency will strengthen
the illusory perception of a response-outcome correlation (assum-
ing that the probability of the desired outcome is high in all cases).
Thus, a mediational hypothesis that transfers motivational or
depression-related differences to a more proximal cause, P(R),
which is amenable to modeling in contingency-learning terms
(unlike the former variables), becomes useful for making specific or
quantitative predictions and modulating the depressive realism
Finally, we would like to discuss the implications of our current
findings for the clinical field. It has been argued that an accurate
and realistic perception of reality is not always a healthy feature
[8,29]. In fact, the illusion of control protects against stressors,
both natural and artificially induced in the laboratory [8].
Therefore, some theories claim that certain optimistic biases and
illusions are attributes of healthy, well-adapted behavior
[29,43,44]. According to this view, the illusion of control may
enhance self-esteem, leading the individual to optimistically believe
that he or she is able to control certain aspects of his or her life
instead of attributing them to chance or to uncontrollable factors.
It has been suggested that depressed people are not as motivated to
protect their self-esteem as healthy people are, leading to the
realistic but uncomfortable perception of uncontrollability [43,44].
Based on the mediational hypothesis presented here, we can
suggest that clinically imposed contingencies aimed at increasing
the depressed patient’s activity and probability of responding (like
those commonly used in cognitive-behavioral therapy) may help
develop these healthy, optimistic illusions. Therapists in the
behavioral tradition are already promoting an increase in patient
activity as a clinical strategy against depression [17,45]. These
therapies are mostly based on theories that conceptualize
depression as a result of a previous history of nonreinforcement,
and being active is normally the best possible way to obtain
reinforcers. We acknowledge the obvious limitations inherent in
working with nonclinical samples, such as college students or
Internet users; still, we note that our results tend to coincide with
the previously mentioned approach to depression treatment, and
they add value to it: Increasing the probability of responding is
good not only because it allows people to obtain controllable
reinforcers but also because it makes their behavior coincide even
with those desired outcomes that occur by mere chance and are
beyond the individual’s control, thus leading to the development of
healthy optimistic illusions.
We would like to thank Esther Calvete and Susana Segura for their very
helpful comments on and discussion of earlier versions of this paper.
Author Contributions
Conceived and designed the experiments: FB HM MAV. Performed the
experiments: FB. Analyzed the data: FB. Wrote the paper: FB HM MAV.
1. Alloy LB, Abramson LY (1979) Judgements of contingency in depressed and
nondepressed students: Sadder but wiser? J Exp Psychol Gen, 108, 441–485.
2. Matute H (1995) Human reactions to uncontrollable outcomes: Further
evidence for superstitions rather than helplessness. Q J Exp Psychol, 48B,
Activity Level and Depressive Realism
PLOS ONE | 7 September 2012 | Volume 7 | Issue 9 | e46203
3. Rudski JM, Lischner MI, Albert LM (1999) Superstitious rule generation is
affected by probability and type of outcome. Psychol Rec, 49, 245–260.
4. Langer EJ (1975) The illusion of control. J Pers Soc Psychol, 32, 311–328.
5. Matute H (1994) Learned helplessness and superstitious behavior as opposite
effects of uncontrollable reinforcement in humans. Learn Motiv, 25, 216–232.
6. Matute H (1996) Illusion of control: Detecting response-outcome independence
in analytic but not in natur alistic conditions. Psychol Sci, 7, 289–293 .
7. Mischel W (1979) On the interface of cognition and personality: Beyond the
person-situation debate. Am Psychol, 34, 740–754. doi:10.1037/0003-
8. Alloy LB, Clements CM (1992) Illusion of Control: Invulnerability to negative
affect and depressive symptoms after laboratory and natural stressors. J Abnorm
Psychol, 101, 234–245. doi:10.1037/0021-843X.101.2.234.
9. Blanco F, Matute H, Vadillo MA (2009) Depressive realism: Wiser or quieter?
Psychol Rec, 59, 551–562.
10. Msetfi RM, Murphy RA, Simpson J (2007) Depressive realism and the effect of
inter-trial-interval on judgments of zero, positive and negative contingencies.
Q J Exp Psychol, 60, 461–481.
11. Msetfi RM, Murphy RA, Simpson J, Kornbrot DE (2005) Depressive realism
and outcome density bias in contingency judgments: The effect of the context
and inter-trial interval. J Exp Psychol Gen, 134, 10–22.
12. Presson PK, Benassi VA (2003) Are depressive symptoms positively or negatively
associated with the illusion of control? Soc Behav Pers, 31, 483–495.
13. Ackermann R, DeRubeis RJ (1991) Is depressive realism real? Clin Psychol Rev,
11, 565–584. doi:10.1016/0272-7358(91)90004-E.
14. Allan LG, Siegel S, Hannah SD (2007) The sad truth about depressive realism.
Q J Exp Psychol, 60, 482–495. doi:10.1080/17470210601002686.
15. Blanco F, Matute H, Vadillo MA (2011) Making the uncontrollable seem
controllable: The role of action in the illusion of control. Q J Exp Psychol, 64,
1290–304. doi:10.1080/17470218.2011.552727.
16. Hannah SD, Beneteau JL (2009) Just tell me what to do: Bringing back
experimenter control in active contingency tasks with the command-perfor-
mance procedure and finding cue density effects along the way. Can J Psychol,
63, 59–73. doi:10.1037/a0013403.
17. Lewinsohn PM, Sullivan JM, Grosscup SJ (1980) Changing reinforcement
events: An approach to the treatment of depression. Psychotherapy-Theor Res
Pract, 17, 322–333. doi:10.1037/h0085929.
18. Pearl J (2009) Causality. Cambridge, UK: Cambridge University Press.
19. Sloman S (2005) Causal Models: How People Think about the World and Its
Alternatives. Oxford, UK: Oxford University Press.
20. Matute H, Vadillo MA, Vegas S, Blanco F (2007) Illusion of contro l in Internet
users and college students. Cyberpsychol Behav, 10, 176–181. doi:10.1089/
21. Frankel MS, Siang S (1999) Ethical and legal aspects of human subjects research
in cyberspace. Report of a workshop convened by the American Association for
the Advancement of Science, Program on Specific Freedom, Responsibility, and
Law, Washington DC. Retrieved May 9, 2005, from
22. Kraut R, Olson J, Banaji M, Bruckman A, Cohen J, et al. (2003) Psychological
Research Online: Opportunities and Challenges. Retrieved May 24, 2012, from
23. Consejo General de Colegios Oficiales de Psico´logos (2010) Co´ digo deontolo´-
gico. Retrieved May 24, 2012, from
24. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961) An inventory for
measuring depression. Arch Gen Psychiatry, 4, 561–571. doi:10.1001/
25. Conde V, Esteban L, Useros E (1976) Adaptacio´ n castellana de la escala de
evaluacio´n conductual para la depresio´ n de Beck. Rev Psicol Gen Apl, 31, 496–
26. Cohen J (1968) Multiple regression as a general data-analytic system. Psychol
Bull, 70, 426–443. doi:10.1037/h0026714.
27. Baron RM, Kenny DA (1986) The moderator-mediator variable distinction in
social psychological research: Conceptual, strategic and statistical considerations.
J Pers Soc Psychol, 51, 1173–1182. doi:10.1037/0022-3514.51.6.1173.
28. Sobel ME (1982) Asymptotic confidence intervals for indirect effects in structural
equation models. In: Leinhardt S, editor. Sociological Methodology). Washing-
ton DC: American Sociological Association. 290–312.
29. Birnbaum MH (2004). Human research and data collection via the Internet.
Annual Review of Psychology, 55, 803–832.
30. Birnbaum MH, Reips U-D (2005). Behavioral research and data collection via
the Internet. In R. W. Proctor & K.-P. L. Vu (Eds.), The handbook of human
factors in Web design (pp. 471–492). Mahwah, NJ: Erlbaum.
31. Alloy LB, Abramson LY (1988) Depressive realism: Four theoretical persp ec-
tives. In: Alloy LB, editor. Cognitive Processes in Depression. New York, NY:
Guilford University Press. 223–265.
32. Dobson KS, Franche RL (1989) A conceptual and empirical review of the
depressive realism hypothesis. Can J Behav Sci, 21, 419–433. doi:10.1037/
33. Moore MT, Fresco DM (2012) Depressive realism: A meta-analytic review. Clin
Psychol Rev, 32, 496–509.
34. Aeschleman SR, Rosen CC, Williams MR (2002) The effect on non-contingent
negative and positive reinforcement operations on the acquisition of superstitious
behaviors. Behav Processes, 61, 37–45. doi:10.1016/S0376-6357(02)00158-4.
35. Benassi VA, Mahler HIM (1985) Contingency Judgments By Depressed College
Students: Sadder But Not Always Wiser. J Pers Soc Psychol, 49, 1323–13 29.
36. Rescorla RA, Wagner AR (1972) A theory of Pavlovian conditioning: Variations
in the effectiveness of reinforcement and nonreinforcement. In: Black AH,
Prokasy WF, editors. Classical conditioning II: Current research and theory.
New York, NY: Appelton-Century-Crofts. 64–99.
37. Matute H, Vadillo MA, Blanco F, Musca SC (2007) Either greedy or well
informed: The reward maximization – unbiased evaluation trade-off. In:
Vosniadou S, Kayser D, Protopapas A, editors. Proceedings of the European
Cognitive Science Conference. Hove, UK: Erlbaum. 341–346.
38. Kao S-F, Wasserman EA (1993) Assessment of an information integration
account of contingency judgment with examination of subjective cell importance
and method of information presentation. J Exp Psychol Learn Mem Cogn, 19,
1363–1386. doi:10.1037/0278-7393.19.6.1363.
39. Wasserman EA, Kao S-F, Van Hamme LJ, Katagari M, Young ME (1996)
Causation and association. In: Shanks DR, Holyoak KJ, Medin DL, editors. The
psychology of learning and motivation, Vol. 34: Causal learning). San Diego,
CA: Academic Press. 207–264.
40. White PA (2002) Causal attribution from covariation information: The
evidential evaluation model. Eur J Soc Psychol, 32, 667–684. doi:10.1002/
41. Wason PC (1960) On the failure to eliminate hypotheses in a conceptual task.
Q J Exp Psychol, 12, 129–140. doi:10.1080/17470216008416717.
42. Nickerson RS (1998) Confirmation bias: A ubiquitous phenomena in many
guises. Rev Gen Psychol, 2, 175–220.
43. Taylor SE, Brown JD (1988) Illusion and well-being: A social psychological
perspective on mental health. Psychol Bull, 103, 192–210. doi:10.1037/0033-
44. Taylor SE, Brown JD (1994) Positive illusion and well-being revisited:
Separating fact from fiction. Psychol Bull, 116, 21–27. doi:10.1037/0033-
45. Lewinsohn PM (1974) A behavioral approach to depression. In: Friedma n RJ,
Katz MM, editors. The psychology of depression: Contemporary theory and
research. Washington, DC: Wiley. 176–178.
Activity Level and Depressive Realism
PLOS ONE | 8 September 2012 | Volume 7 | Issue 9 | e46203
... Another approach is to assume that perceived control is data-driven. Thus, it has been argued that instead of, or perhaps in addition to, pre-existing biases being the main influencer of perceived control ratings, it is the processing of and learning about specific aspects of contingency experience (Blanco, Matute, & Vadillo, 2011;Blanco, Matute, & Vadillo, 2012;Blanco et al., 2009;Byrom, Msetfi, & Murphy, 2015;Msetfi, Wade, & Murphy, 2013) or the current motivational state of the participant (e.g., Baker, Msetfi, Hanley, & Murphy, 2011). For example, Matute (1996) has argued that rates of behaviour, affect the degree of control perceived. ...
... Although participants might be instructed to sample actions and non-actions, there are variations in behaviour. Matute argued that non-depressed people are more likely to do this, and indeed there is evidence consistent with this position (Blanco et al., 2009(Blanco et al., , 2012. ...
... This type of difference may make people showing signs of depression to be more sensitive to non-controllable adverse events (Johnson & Sarason, 1978). However, our evidence while correlational, suggests that the High BDI participants: a pre-existing context attentional bias produced more responses ; or a general increased response rate enhanced overall control judgments, (Blanco et al., 2012), particularly in High BDI participants where they over-integrate information from the context . ...
Depression has been linked to weakened perceptions of control. The experimental evidence derives from tasks with exposure to stable action-outcome contingencies. One assumption has been that performance represents a general cognitive bias that might manifest itself by a global performance difference. Another view is that people have specific situational perceptions of control reflecting their recent actions and the contingencies to which they are currently experiencing. In an experiment with N = 179, participants acquired one of four action-outcome sequences (Constant or Variable). We measured how learning was reflected in ratings of control and probability of responding in relation to mood. In three experimental treatments, the overall contingency across training involved an average moderate degree of control (ΔP = 0.25), but differed in how control varied (Constant or one of two Variable treatments). A fourth, control treatment involved a Constant zero degree of control (ΔP = 0.00). Participants rated their control before, during and after each sequence, providing measures of pre-existing bias, ratings of control in specific situations and generalised control perceptions. Specific control ratings were only influenced by the contingency experience and not pre-existing bias. Higher scores on the Beck’s depression inventory were associated with weakened association between action and context ratings. Overall, these data suggest that human agency is related to rates of responding and that mood is related to a difference in sensitivity to the ratings of and responding to the context.
... Исследование Alloy и Abramson (1979) показало, что люди, страдающие депрессией, более точны, в обнаружении их отсутствия контроля в условных нулевой сопряженности причины и результата, чем люди, не страдающие депрессией. Согласно Blanco F. et al. (2009Blanco F. et al. ( , 2012, одним из аспектов депрессии является большая пассивность, то есть снижение способности инициировать добровольные ответы. Предполагается что испытуемые без депрессии действуют с большей частотой, чтобы получить результат. ...
... В рамках «эвристической модели» существуют также несколько подходов к пониманию иллюзии контроля: в социальной психологии этот феномен воспринимается как мотивационное искажение или искажение самозащиты Weiner B.,1979). В некоторых работах исследователи предполагают этот феномен результатом ассоциативного обучения (Blanco F. et al., 2012;Matute H., 1996), а в более поздних как искажение восприятия сопряженности (Blanco F., 2017;Matute H. et al., 2015 ...
... Как уже упоминалось при описании связи иллюзии контроля и настроения, исследование Alloy и Abramson (1979) (Blanco et al., 2009(Blanco et al., , 2012. Эти черты связаны с дисфорией, но не обязательно подразумевают причину и следствие. ...
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.
... To test this hypothesis, we administer measures of perceived control throughout the task and also at the end. Second, given that motor response rates profoundly influence beliefs about control and are often diminished among those with depression (Blanco et al., 2012), we measure whether the frequency of responding helps explain how depression relates to beliefs about control (Blanco et al., 2009). Finally, we ask participants to estimate outcome probabilities based on both possible responses (pressing and not pressing the button), since true contingencies exist for both. ...
... When we considered more frequent ratings of subjective control, we still found no association between depressive symptoms with outcome bias. Past work has suggested that psychomotor retardation in depression could be one mechanism through which depressive symptoms contribute to lower illusory control (Blanco et al., 2012), but we found no support for this hypothesis when examining frequency of responding in our samples. ...
Full-text available
The theory of depressive realism holds that depressed individuals are less prone to optimistic bias, and are thus more realistic, in assessing their control or performance. Since the theory was proposed 40 years ago, many innovations have been validated for testing cognitive accuracy, including improved measures of bias in perceived control and performance. We incorporate several of those innovations in a well-powered, pre-registered study designed to identify depressive realism. Amazon MTurk workers (N = 246) and undergraduate students (N = 134) completed a classic contingency task, an overconfidence task, and measures of mental health constructs, including depression and anxiety. We measured perceived control throughout the contingency task, allowing us to compare control estimates at the trial-level to estimates assessed at task conclusion. We found no evidence that depressive symptoms relate to illusory control or to overconfidence. Our results suggest that despite its popular acceptance, depressive realism is not replicable.
... The causal illusion and related biases have been proposed to underlie many societal problems including social prejudice and stereotype formation (Hamilton and Gifford, 1976;Kutzner et al., 2011;Matute et al., 2015). Importantly, the causal illusion has been detected in healthy participants (it is not the result of a disordered or ill-adapted mind), and it even shows variations depending on factors that affect the general population, such as mood (Alloy and Abramson, 1979;Msetfi et al., 2005;Blanco et al., 2012). In general, this illusion is considered a basic feature of the way normal cognitive systems work, rather than an anomaly. ...
... At the same time, the illusion has also been associated to well-being and optimism, as it is less prevalent in depressed people (Taylor and Brown, 1988;Blanco, 2017). Previous research has identified various factors that modulate this bias from the observed frequencies of the events involved (Allan and Jenkins, 1983;Buehner and May, 2004) to individual differences such as mood (Alloy and Abramson, 1979;Msetfi et al., 2005;Blanco et al., 2012). Yet another factor that could modulate the illusion is previous knowledge and expectations . ...
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).
... An agent's action produces data for the agent, which they use to form beliefs about their agency [31]. Indeed, a link between probability of acting and objective contingency has been established in free-operant tasks [32]- [34]. Moreover, agency may emerge from a form of inferential reasoning [18], [35], [36], wherein agents not only rely on direct sensory input and statistical metrics, but also engage in processes that involve learning about the dynamics and causes of the latent states of the world. ...
Full-text available
Agential learning refers to the process of forming beliefs regarding one’s degree of control over actions and outcomes in their environment. We first provide an overview and evaluation of associative, statistical, and Bayesian models of agential learning. We then argue that the existing models have limitations in explaining the process of agential learning. Finally, we introduce an active inference account of agential learning, and present results from simulations. We propose that the active inference framework may provide a comprehensive model of agential learning describing three fundamental processes: (i) perception, (ii) learning, and (iii) action.
... In the depressed view (so-called "depressive realism"), the reward/punishment ratio is evenly balanced, which substantially reduces opportunities for learning and ultimately makes any action pointless (Nettle and Bateson, 2012). This account agrees with clinical reports of greater passivity in depressed people, that is, a reduced ability to initiate voluntary actions (Blanco, Matute, & Vadillo, 2012). Acting with less frequency would make depressed individuals exposed to fewer incidental associations between actions and action-contingent events (reduced "action-density" bias, see Matute et al., 2015), which might in turn impede learning instrumental contingencies and aggravate depressive symptoms in the long run. ...
Full-text available
Most people envision themselves as operant agents endowed with the capacity to bring about changes in the outside world. This ability to monitor one's own causal power has long been suggested to rest upon a specific model of causal inference, i.e., a model of how our actions causally relate to their consequences. What this model is and how it may explain departures from optimal inference, e.g., illusory control and self-attribution biases, are still conjecture. To address this question, we designed a series of novel experiments requiring participants to continuously monitor their causal influence over the task environment by discriminating changes that were caused by their own actions from changes that were not. Comparing different models of choice, we found that participants' behaviour was best explained by a model deriving the consequences of the forgone action from the current action that was taken and assuming relative divergence between both. Importantly, this model agrees with the intuitive way of construing causal power as "difference-making" in which causally efficacious actions are actions that make a difference to the world. We suggest that our model outperformed all competitors because it closely mirrors people's belief in their causal power - a belief that is well-suited to learning action-outcome associations in controllable environments. We speculate that this belief may be part of the reason why reflecting upon one's own causal power fundamentally differs from reasoning about external causes.
... Elucidation of the illusion of control that emphasizes the role of coincidences between behavior and environmental changes is an important step toward providing a basic background for understanding behavioral and learning mechanisms that are related to the origins of false beliefs (Blanco, 2017;Blanco et al., 2009Blanco et al., , 2011Blanco et al., , 2012Blanco et al., , 2013Matute, 1996;Matute et al., 2007). The present data support this approach to better understand the general notion of the illusion of control. ...
Full-text available
The notion of superstitious behavior can provide a basic background for understanding such notions as illusions and beliefs. The present study investigated the social mechanism of the transmission of superstitious behavior in an experiment that utilized participant replacement. The sample was composed of a total of 38 participants. Participants performed a task on a computer: they could click a colored rectangle using the mouse. When the rectangle was in a particular color, the participants received points independently of their behavior (variable time schedule). When the color of the rectangle was changed, no points were presented (extinction). Under an Individual Exposure condition, ten participants worked alone on the task. Other participants were exposed to the same experimental task under a Social Exposure condition, in which each participant first learned by observation and then worked on the task in a participant replacement (chain) procedure. The first participant in each chain in the Social Exposure condition was a confederate who worked on the task “superstitiously,” clicking the rectangle when points were presented. Superstitious responding was transmitted because of the behavior of the confederate. This also influenced estimates of personal control. These findings suggest that social learning can facilitate the acquisition and maintenance of superstitious behavior and the illusion of control. Our data also suggest that superstitious behavior and the illusion of control may involve similar learning principles.
Previous research indicates that economic scarcity affects people's judgments, decisions, and cognition in a variety of contexts, and with various consequences. We hypothesized that scarcity could sometimes reduce cognitive biases. Specifically, it could reduce the causal illusion, a cognitive bias that is at the heart of superstitions and irrational thoughts, and consists of believing that two events are causally connected when they are not. In three experiments, participants played the role of doctors deciding whether to administer a drug to a series of patients. The drug was ineffective, because the percentage of patients recovering was identical regardless of whether they took the drug. We manipulated the budget available to buy the drugs, tough all participants had enough for all their patients. Even so, participants in the scarce group reduced the use of the drug and showed a lower causal illusion than participants in the wealthy group. Experiments 2 and 3 added a phase in which the budget changed. Participants who transitioned from scarcity to wealth exhibited a reduced use of resources and a lower causal illusion, whereas participants transitioning from wealth to scarcity were unaffected by their previous history.
Full-text available
Many experiments have shown that humans and other animals can detect contingency between events accurately. This learning is used to make predictions and to infer causal relationships, both of which are critical for survival. Under certain conditions, however, people tend to overestimate a null contingency. We argue that a successful theory of contingency learning should explain both results. The main purpose of the present review is to assess whether cue-outcome associations might provide the common underlying mechanism that would allow us to explain both accurate and biased contingency learning. In addition, we discuss whether associations can also account for causal learning. After providing a brief description on both accurate and biased contingency judgments, we elaborate on the main predictions of associative models and describe some supporting evidence. Then, we discuss a number of findings in the literature that, although conducted with a different purpose and in different areas of research, can also be regarded as supportive of the associative framework. Finally, we discuss some problems with the associative view and discuss some alternative proposals as well as some of the areas of current debate. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
In the last decades, cognitive Psychology has provided researchers with a powerful background and the rigor of experimental methods to better understand why so many people believe in pseudoscience, paranormal phenomena and superstitions. According to recent evidence, those irrational beliefs could be the unintended result of how the mind evolved to use heuristics and reach conclusions based on scarce and incomplete data. Thus, we present visual illusions as a parallel to the type of fast and frugal cognitive bias that underlies pseudoscientific belief. In particular, we focus on the causal illusion, which consists of people believing that there is a causal link between two events that coincide just by chance. The extant psychological theories that can account for this causal illusion are described, as well as the factors that are able to modulate the bias. We also discuss that causal illusions are adaptive under some circumstances, although they often lead to utterly wrong beliefs. Finally, we mention several debiasing strategies that have been proved effective in fighting the causal illusion and preventing some of its consequences, such as pseudoscientific belief.
Full-text available
In this article, we attempt to distinguish between the properties of moderator and mediator variables at a number of levels. First, we seek to make theorists and researchers aware of the importance of not using the terms moderator and mediator interchangeably by carefully elaborating, both conceptually and strategically, the many ways in which moderators and mediators differ. We then go beyond this largely pedagogical function and delineate the conceptual and strategic implications of making use of such distinctions with regard to a wide range of phenomena, including control and stress, attitudes, and personality traits. We also provide a specific compendium of analytic procedures appropriate for making the most effective use of the moderator and mediator distinction, both separately and in terms of a broader causal system that includes both moderators and mediators. (46 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Full-text available
Confirmation bias, as the term is typically used in the psychological literature, connotes the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations, or a hypothesis in hand. The author reviews evidence of such a bias in a variety of guises and gives examples of its operation in several practical contexts. Possible explanations are considered, and the question of its utility or disutility is discussed.
Full-text available
We examined whether individual differences in susceptibility to the illusion of control predicted differential vulnerability to depressive responses after a laboratory failure and naturally occurring life stressors. The illusion of control decreased the likelihood that subjects (N= 145) would (a)show immediate negative mood reactions to the laboratory failure, (b) become discouraged after naturally occurring negative life events, and (c) experience increases in depressive symptoms a month later given the occurrence of a high number of negative life events. In addition, the stress-moderating effect of the illusion of control on later depressive symptoms appeared to be mediated in part by its effect on reducing the discouragement subjects experienced from the occurrence of negative life events. These findings provide support for the hopelessness theory of depression and for the optimistic illusion-mental health link.
Full-text available
Recent research has shown superstitious behaviour and illusion of control in human subjects exposed to the negative reinforcement conditions that are traditionally assumed to lead to the opposite outcome (i.e. learned helplessness). The experiments reported in this paper test the generality of these effects in two different tasks and under different conditions of percentage (75% vs. 25%) and distribution (random vs. last-trials) of negative reinforcement (escape from uncontrollable noise). All three experiments obtained superstitious behaviour and illusion of control and question the generality of learned helplessness as a consequence of exposing humans to uncontrollable outcomes.
Full-text available
Confirmation bias, as the term is typically used in the psychological literature, connotes the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations, or a hypothesis in hand. The author reviews evidence of such a bias in a variety of guises and gives examples of its operation in several practical contexts. Possible explanations are considered, and the question of its utility or disutility is discussed. When men wish to construct or support a theory, how they torture facts into their service! (Mackay, 1852/ 1932, p. 552) Confirmation bias is perhaps the best known and most widely accepted notion of inferential error to come out of the literature on human reasoning. (Evans, 1989, p. 41) If one were to attempt to identify a single problematic aspect of human reasoning that deserves attention above all others, the confirma- tion bias would have to be among the candidates for consideration. Many have written about this bias, and it appears to be sufficiently strong and pervasive that one is led to wonder whether the bias, by itself, might account for a significant fraction of the disputes, altercations, and misun- derstandings that occur among individuals, groups, and nations.
One hundred and fifty participants played a computer task in which points were either gained (reinforcement) or lost (punishment) randomly on 75%, 50%, or 25% of trials. Despite the noncontingent nature of the task, participants frequently suggested superstitious rules by which points were either gained or lost. Rules were more likely to be suggested and supported higher confidence ratings under conditions of maximal reinforcement or minimal punishment, and participants gaining points tended to express more rules than did those losing points. Superstitious rule generation was in no way related to a person's locus of control, as measured by Rotter's Internal-External Scale. Participants losing points were more accurate in keeping track of their total number of points than were participants gaining points. Results are discussed in terms of reinforcement and punishment's effects on the stimulus control of rule-governed behavior, and comparisons are drawn with the illusion of control and learned helplessness literature.
Human beings are active agents who can think. To understand how thought serves action requires understanding how people conceive of the relation between cause and effect, between action and outcome. This book presents the question, in cognitive terms: how do people construct and reason with the causal models we use to represent our world? A revolution is occurring in how statisticians, philosophers, and computer scientists answer this question. Those fields have ushered in new insights about causal models by thinking about how to represent causal structure mathematically, in a framework that uses graphs and probability theory to develop what are called causal Bayesian networks. The framework starts with the idea that the purpose of causal structure is to understand and predict the effects of intervention. How does intervening on one thing affect other things? This is not a question merely about probability (or logic), but about action. The framework offers a new understanding of mind: thought is about the effects of intervention and cognition is thus intimately tied to actions that take place either in the actual physical world or in imagination, in counterfactual worlds. This book offers a conceptual introduction to the key mathematical ideas, presenting them in a non-technical way, focusing on the intuitions rather than the theorems. It tries to show why the ideas are important to understanding how people explain things and why thinking not only about the world as it is but the world as it could be is so central to human action. The book reviews the role of causality, causal models, and intervention in the basic human cognitive functions: decision making, reasoning, judgment, categorization, inductive inference, language, and learning. In short, the book offers a discussion about how people think, talk, learn, and explain things in causal terms, in terms of action and manipulation.
The difficulties inherent in obtaining consistent and adequate diagnoses for the purposes of research and therapy have been pointed out by a number of authors. Pasamanick12 in a recent article viewed the low interclinician agreement on diagnosis as an indictment of the present state of psychiatry and called for "the development of objective, measurable and verifiable criteria of classification based not on personal or parochial considerations, but on behavioral and other objectively measurable manifestations."Attempts by other investigators to subject clinical observations and judgments to objective measurement have resulted in a wide variety of psychiatric rating scales.4,15 These have been well summarized in a review article by Lorr11 on "Rating Scales and Check Lists for the Evaluation of Psychopathology." In the area of psychological testing, a variety of paper-and-pencil tests have been devised for the purpose of measuring specific