Orosz Et Al (2007)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Behavioural Brain Research 180 (2007) 13

Research report

Repeated measurements of learned irrelevance by


a novel within-subject paradigm in humans
Ariane Orosz a,b,e, , Joram Feldon a , Gilad Gal c ,
Andor Simon d , Katja Cattapan-Ludewig b,e
a

Laboratory of Behavioral Biology, ETH Zurich, Switzerland


b University Hospital of Psychiatry, Bern, Switzerland
c Mental Health Epidemiology and Psychosocial Aspects of Illness, The Gertner Institute for Epidemiology and
Health Policy Research, Sheba Medical Center, Tel Hashomer, Israel
d Specialised Outpatient Service for Early Psychosis, Department of Psychiatry, Bruderholz, Switzerland
e MediQ, Psychiatric Services of Aargau Canton AG, Research Department, Brugg, Switzerland
Received 20 September 2006; received in revised form 1 February 2007; accepted 6 February 2007
Available online 12 February 2007

Abstract
Learned irrelevance (LIrr) refers to the retardation of classical conditioning following preexposure of the to-be-associated stimuli. Healthy
volunteers have been tested on three occasions with a new LIrr paradigm avoiding methodological problems which afflict traditional paradigms.
A significant LIrr effect was demonstrated on each occasion. Thus, the new paradigm enables repeated measurements of LIrr and might be useful
in evaluating long-term effects of medication in psychiatric disorders exhibiting aberrant LIrr.
2007 Elsevier B.V. All rights reserved.
Keywords: Learned irrelevance; Latent inhibition; Repeated measurements; Within-subject design; Information processing; Attention; Schizophrenia

Learned irrelevance is considered a measure of information processing, specifically, the ability of normal individuals to
ignore irrelevant internal and external stimuli in order to protect
themselves from stimulus overload. Learned irrelevance (LIrr)
refers to the retardation of associative learning, such as classical
conditioning - which stands for the formation of an association between a conditioned stimulus (CS) and an unconditioned
stimulus (US) - following repeated unpaired presentations of
the CS and the US. LIrr is closely related to the to date more
extensively studied phenomenon of latent inhibition (LI) [9]. LI
differs from LIrr in terms of the preexposure stage: in an LI procedure only the CS is preexposed, while in LIrr both the CS and
the US are presented inconsequentially prior to conditioning.
LIrr has been shown to produce comparable data to LI as both
phenomena were observed to be deficient in acute schizophrenia
patients ([1], for a review see [5]). Therefore, disrupted LI and

Corresponding author at: University Hospital of Psychiatry (UPD),


Department of Psychiatric Neurophysiology, Murtenstr. 21, CH-3010 Bern,
Switzerland. Tel.: +41 31 632 88 92; fax: +41 31 632 89 44.
E-mail address: ariane.orosz@insel.ch (A. Orosz).

0166-4328/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbr.2007.02.008

LI-like phenomena have become an important tool to characterize cognitive and attentional deficits in schizophrenia and might
even be treated as a state marker of it [10]. Disrupted or reduced
LI in schizophrenia patients takes the form of faster learning
of the CSUS association compared to normal subjects. This
implies that patients who show a general cognitive performance
deficit actually perform better in the LI task following preexposure than control subjects [1,5]. Thus, factors such as general
deficits in intellectual ability, medication side effects or lack of
motivation of schizophrenia patients can be excluded from being
responsible for LI disruption [5,6].
Traditional LI paradigms used to date are unsuitable for
repeated measurements [1,3]. This is due to the fact that traditional LI paradigms use an instrumental learning task and the
dependent measure is the number of trials needed to learn the
CSUS association [1,4]. As these associations are quite simple, learning retardation by CS preexposure occurs at best only
on one occasion, i.e. no LI effect can be repeatedly observed in
the same subject, as the association has already been learned.
An additional limitation to these traditional LI paradigms is that
they apply a between-subject design. Thus, an LI effect can be
determined only by group comparisons [6].

A. Orosz et al. / Behavioural Brain Research 180 (2007) 13

Some advanced LI paradigms [7,8] and the LIrr paradigm


used in the present study enable a within-subject design as they
involve more complex learning tasks which are not based on
a simple solution. Moreover, they measure the reaction time
(RT) as the dependent variable. It is suggested that by applying
a complex task and measuring RTs repeated measurements of
LIrr can be performed within one individual.
Sixteen healthy volunteers (3f, 13m) with a mean age of 23.4
years (S.D. = 2, range = 2029) were tested on three occasions
with the new LIrr paradigm at monthly intervals. A diagnostic
interview (DIA-X [11]) was performed with the subjects prior
to the first test session in order to confirm the absence of a
personal or family history of psychiatric disorders. The study
was carried out at the Psychiatric Services of Aargau Canton,
Switzerland. Subjects were recruited by electronic advertisements and by placards. The study protocol and consent forms
were reviewed and approved by the Ethical Committee of the
Psychiatric Services of Aargau Canton. The LIrr test used in the
present study is a simple computerized visual target-detecting
task of 7.5 min duration. It is an advanced version of the LIrr
paradigm developed by Young et al. [12] which was further
modified by Gal et al. [2]. The paradigm appeared as a RT task
in which the subjects were instructed to press the space bar of the
computer keyboard as soon as the letter X, which represented
the target, appeared on the screen. In addition to the target letter,
there were 10 different capital Latin letters (vowels: A, E, I, O, U
and five consonants: B, D, T, Y and Z) occurring in the test. All
characters were yellow and were presented on blue background.
The letters were all of the same size and were positioned in the
center of the screen. They appeared continuously one after the
other in a 1 s rate. A test session was composed of 75 targets
and 375 non-target letters, thus a total of 450 letters. As each
letter persisted for 1 s on the screen, the duration of a session was
450 s. The non-target letters were divided into two groups, nonpreexposed (NPE) and preexposed (PE). The five vowels (A,
E, I, O and U) were assigned to be the NPE letters, while five
consonants (B, D, T, Y and Z) constituted the PE letter group.
NPE letters, PE letters and the targets were presented according to the schedules of three different conditions: preexposed
(PE), non-preexposed (NPE) and random (R). Each condition
was segmented in five blocks. The total of 15 blocks of the three
different conditions were always presented in the same order,
whereby the first block of a test session was always an R block
and there were never two successive blocks of the same condition. All blocks contained 30 letters: 5 targets, 5 target predictor
letters which are presented immediately before the target, and
20 filler letters. Filler letters consisted of characters of the PE
group (consonants). They served to fill the period between the
predictor letter-target contingencies (e.g. B-X in a PE and O-X
in a NPE block). There were one to eight, on average four, filler
letters between the predictor letter-target contingencies. Filler
letters never predicted the target.
In an NPE condition block, the target was predicted five consecutive times by the same vowel, e.g. A. Thus, in each of the five
NPE blocks one of the five NPE letters (vowels) was used as the
target predictor. Consequently, the five vowels were distributed
to the five NPE condition blocks.

It is important to note that a particular vowel is only presented


in the NPE block in which it predicted the target. It does not
appear before and will not be presented later in the test session.
In the PE condition, the PE letters served as target predictors.
Similarly to NPE, in a PE block the target was preceded five
times by the same PE letter, e.g. five times by B. Thus, each
consonant acted as target predictor in one particular PE condition
block. In the R blocks the five targets appeared once after each of
the five PE letters, i.e. five times by a different consonant (once
after B, once after D, etc.). Hence, preexposure to the PE letters
took place especially in the R blocks as well as during filler
letter presentations, where the subjects learned that a consonant
was not necessarily followed by the target. In contrast, a vowel
reliably predicted a target. Therefore, in the NPE blocks there
was full prediction of the target, while in PE the target could only
partly be predicted and in R not at all. Accordingly, the RT to the
target was expected to be the lowest (fastest) in NPE, higher in
PE and the highest (slowest) in R. In case of LIrr, performance
on the RT task is supposed to be significantly faster in the NPE
than in the PE condition.
Prior to the test, the subjects were instructed that they were
going to perform a RT task which lasted about 7 min. During
this time they should watch the screen and whenever the letter
X appeared, they had to press the space bar as quickly as they
could. Moreover, they were also told to attentively follow the
non-target letters as they may give a hint to the appearance of
the X. Finally, they were asked to avoid making mistakes.
The RTs, i.e. the time (in ms) between the target onset and the
pressing of the space bar was recorded as the dependent variable.
To be included in the analysis, subjects had to detect at least
60 of the 75 targets in a test session. In the present study, this
criterion has been met by all subjects. RTs were analyzed using
3 3 5 repeated measures analysis of variance (ANOVA) with
test session (1st, 2nd, 3rd), condition (R, PE, NPE) and block
(15) as within-subject factors.
There was no effect of session, as the performance in the
three different conditions, especially R and PE, remained stable
(similar RTs) over sessions (Fig. 1). The effect of condition was
significant (F(2.30) = 31.91, p < 0.001). This effect was, on the
one hand, due to a difference in performance in the PE compared
with the NPE condition, which was either significant or close to
significant, reflecting an LIrr effect in all three test sessions [post
hoc tests (Tukeys HSD) revealed for baseline: p = 0.09; 2nd test
session: p < 0.001; 3rd test session: p < 0.001]. On the other hand,
the condition effect reflected higher RTs in R compared to both
PE and NPE (baseline-R versus PE: p < 0.05, R versus NPE:
p < 0.001; 2nd test session-R versus PE: p < 0.01, R versus NPE:
p < 0.001; 3rd test session-R versus PE: p < 0.05, R versus NPE:
p < 0.001). As expected, slowest performance was evident in the
R condition. The interaction between session and condition was
significant (F(4.60) = 7.55, p < 0.001). Post hoc tests showed that
the significant interaction was due to a decrease in RT in the NPE
condition while there were hardly any changes in performance
in R and PE conditions.
LIrr and LI are defined as the reduced associative learning
when preexposed stimuli, i.e. PECS (consonants) are involved.
Disruption of LI, which is the case when the same subject is

A. Orosz et al. / Behavioural Brain Research 180 (2007) 13

which enables repeated measurement of a LIrr effect is its usefulness for long-term studies. LIrr disruption is considered as an
indication for dysfunctional attentional information processing
as it is present in acute schizophrenia. Repeated measurements
of LIrr in schizophrenia patients might be helpful in evaluating
the development of attentional dysfunctions over the course of
disease or in observing the long-term effect of new antipsychotic
drugs on attentional information processing deficits.
Acknowledgments
Financial support was provided by the Sarasota Opera Benefit
through a 2003 NARSAD Young Investigators Award received
by Katja Cattapan-Ludewig. The authors are specially grateful
for the assistance of Pietro Ballinari in statistical matters.
Fig. 1. The figure depicts the average reaction times in the three different conditions (R, PE and NPE) on three occasions. Independent of the test session, the
slowest performance occurred in R, where subjects showed significantly higher
reaction times compared to PE and NPE. The subjects showed a significant LIrr
effect in all test sessions in terms of significantly (or trend to) slower reaction
times in PE vs. NPE. While performance in R and PE remained stable, reaction
times in NPE decreased over sessions (from 1st to 2nd session: p < 0.001, from
2nd to 3rd session: n.s.). The whiskers represent the standard error of the mean.
# p = 0.09 (trend); : p < 0.05; * p < 0.001.

repeatedly tested, is supposed to occur because the subject starts


to learn the PECS-target contingency after a specific number
of trials. In the present study subjects have exhibited similar
RTs in the PE blocks across sessions. Moreover, they showed
reduced learning on the PECS compared to the learning on
the NPECSs in all three test sessions. These results indicate
that a stable LIrr effect could be repeatedly measured on three
occasions.
It is noticeable that the LIrr effect seems to strengthen over
sessions. Actually, this improvement in LIrr is caused by the
subjects learning on the NPECSs. The learning on NPECSs is
facilitated, as the NPE-letters (vowels) always correctly predict
the target. Thus, in NPE there is basically a classical conditioning
procedure, in which the subject learns that a vowel (CS) is a
signal for the target (US).
To our knowledge this is the first study examining repeated
measurements with the novel within-subject LIrr test. Furthermore, no comparable study has been reported in the human LI
literature. We were able to show that by applying a complex
task in which the CSUS association can be learned only rudimentarily, a LIrr effect can be elicited several times in the same
subject. The major advantage of such a within-subject LIrr test

References
[1] Baruch I, Hemsley DR, Gray JA. Differential performance of acute
and chronic schizophrenics in a latent inhibition task. J Nerv Ment Dis
1988;176:598606.
[2] Gal G, Mendlovic S, Bloch Y, Beitler G, Levkovitz Y, Young AM, Feldon J,
Ratzoni G. Learned irrelevance is disrupted in first-episode but not chronic
schizophrenia patients. Behav Brain Res 2005;159:26775.
[3] Ginton A, Urca G, Lubow RE. The effects of preexposure to a nonattended stimulus on subsequent learning: latent inhibition in adults. Bull
Psychonomic Soc 1975;5:58.
[4] Gray NS, Hemsley DR, Gray JA. Abolition of latent inhibition in acute,
but not chronic schizophrenia. Neurol Psychiatr Brain Res 1992;1:839.
[5] Gray NS, Snowden RJ. The relevance of irrelevance to schizophrenia.
Neurosci Biobehav Rev 2005;29:98999.
[6] Gray NS, Snowden RJ, Peoples M, Hemsley DR, Gray JA. A demonstration of within-subjects latent inhibition in the human: limitations and
advantages. Behav Brain Res 2003;138:18.
[7] Lubow RE, De la CG. Latent inhibition as a function of schizotypality and
gender: implications for schizophrenia. Biol Psychol 2002;59:6986.
[8] Lubow RE, Kaplan O, Abramovich P, Rudnick A, Laor N. Visual search in
schizophrenia: latent inhibition and novel pop-out effects. Schizophr Res
2000;45:14556.
[9] Lubow RE, Moore AU. Latent inhibition: the effect of nonreinforced
pre-exposure to the conditional stimulus. J Comp Physiol Psychol
1959;52:4159.
[10] Swerdlow NR, Stephany N, Wasserman LC, Talledo J, Sharp R, Auerbach PP. Dopamine agonists disrupt visual latent inhibition in normal
males using a within-subject paradigm. Psychopharmacology (Berl)
2003;169:31420.
[11] Wittchen HU, Pfister H. DIA-X interview. Frankfurt: Swets Test; 1997.
[12] Young AM, Kumari V, Mehrotra R, Hemsley DR, Andrew C, Sharma
T, Williams SC, Gray JA. Disruption of learned irrelevance in acute
schizophrenia in a novel continuous within-subject paradigm suitable for
fMRI. Behav Brain Res 2005;156:27788.

You might also like