Behavioural Brain Research 133 (2002) 333 /342
www.elsevier.com/locate/bbr
Research report
The partial reinforcement extinction effect in humans: effects of
schizophrenia, schizotypy and low doses of amphetamine
Nicola S. Gray a,b,*, Alan D. Pickering c, Robert J. Snowden a, David R. Hemsley d,
Jeffrey A. Gray d
a
b
School of Psychology, Cardiff University, PO Box 901, Park Place, Cardiff CF10 3YG, South Wales, UK
South Wales Forensic Psychiatric Service at Caswell Clinic, Glanrhyd Hospital, Bridgend CF31 4LN, UK
c
Department of Psychology, Goldsmith’s College, London, UK
d
Department of Psychology, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, UK
Received 29 October 2001; received in revised form 14 January 2002; accepted 14 January 2002
Abstract
The partial reinforcement extinction effect (PREE) was studied in human subjects. It has been suggested that the PREE depends
on neural mechanisms critical to the cognitive dysfunction which underlines acute schizophrenia. We therefore predicted that the
PREE should be reduced, through decreased resistance to extinction in the partial reinforcement (PR) condition, in various types of
individual: (a) healthy volunteers given low doses of oral amphetamine; (b) those in the acute (but not chronic) phase of a
schizophrenic illness and; (c) healthy volunteers with high scores on personality measures of schizotypy. Despite obtaining robust
demonstrations of PREE in all experiments, none of these predictions were confirmed. A single, low dose, of amphetamine had no
effect on either continuous reinforcement (CR) or partial reinforcement (PR). Acute and chronic schizophrenic patients showed a
reduced PREE compared to controls. However this was due to increased resistance to extinction in the CR groups. Finally, high
schizotypy scores were associated with greater PREE, attributable to both decreased extinction in the CR condition and increased
extinction in the PR condition. The results of these experiments on human PREE provide no support that PREE is a valid paradigm
with which to explore the cognitive dysfunction underlying schizophrenia. # 2002 Elsevier Science B.V. All rights reserved.
Keywords: Partial reinforcement extinction effect; Schizophrenia; Schizotypy; Amphetamine
1. Introduction
Gray, Feldon, Rawlins, Hemsley and Smith [20] have
provided a putative model of the neuropsychology of
acute schizophrenia. The model draws together evidence
of neurochemical and neuroanatomical abnormalities in
schizophrenia, with evidence of cognitive deficits
thought to underpin symptom formation. The model
has three central tenets [19,20]. (1) It puts forward a
hypothesis that the crucial substrate of the positive
symptoms of acute schizophrenia lies in an abnormality
in the projections from the septo-hippocampal system
(via the subiculum and entorhinal cortex) to nucleus
accumbens [33]. (2) This abnormality is proposed to
* Corresponding author. Tel: 44-29-2087-6259; fax: 44-292087-4858
E-mail address: grayns@cardiff.ac.uk (N.S. Gray).
interact with the ascending dopaminergic projection to
nucleus accumbens with consequences functionally
equivalent to increased dopaminergic activity in the
mesolimbic dopamine system. (3) This neurochemical
imbalance is proposed to impair the normal, contextdependent inhibition of attention to redundant stimuli,
that is hypothesised by the model to be the core
cognitive deficit underlying the positive symptoms of
acute schizophrenia. Gray et al. [20] proposed that the
cognitive inability of acute schizophrenic patients to
screen irrelevant stimuli from awareness, and thus to be
able to focus attentional resources upon relevant
information in order to learn about contingencies within
the environment, produced the symptomatic profile of
acute schizophrenia (see also [19]).
The model was based upon several lines of empirical
evidence from work with animals. First, a variety of
sources of evidence had suggested that the nucleus
0166-4328/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 6 - 4 3 2 8 ( 0 2 ) 0 0 0 1 9 - 0
334
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
accumbens acts as an interface between the limbic
system and the basal ganglia [41]. Evidence from
experimental work with rats was used to propose the
hypothesis that the limbic system discharges a general
comparator function and compares, on a moment-bymoment basis, the current state of the perceptual world
with a predicted state [18]. Working in synchrony with
the limbic system, the basal ganglia discharges the
general function of motor programming. These neuroanatomical functions were thus proposed by the Gray et
al. [20] model to underpin the psychological mechanism
of distinguishing relevant from irrelevant stimuli, and
acting upon only the relevant. Second, a series of
experiments with rats, using three key behavioural
paradigms, had demonstrated a similar disruption in
cognitive processing (characterised by an inability to
screen out irrelevant information from the control of
behaviour) by both the administration of the indirect
dopamine agonist amphetamine, and damage to the
hippocampal formation.
The experimental paradigms used in these animal
experiments were latent inhibition (LI), the Kamin
blocking effect (KBE; otherwise termed conditioned
blocking) and the partial reinforcement extinction effect
(PREE). LI refers to a retardation of learning the
significance of a conditioned stimulus if it has previously
been pre-exposed without consequence [37]. KBE refers
to the decreased ability of a subject to form a conditioned association between a stimulus and an event if
that subject has formed a previous conditioned association between another stimulus and the same event and
the two predictive stimuli are presented concurrently
[32]. PREE refers to an increased resistance to extinction
following partial reinforcement (PR; rewarded and nonrewarded trials randomly interspersed) as compared to
continuous reinforcement (CR; all trials rewarded).
Behavioural responding persists far longer in the PR
group than the CR group when all rewards cease
(extinction). This resistance to extinction in the PR
group may be interpreted to reflect that the animals
have learnt to ignore non-reward in a manner analogous
to how animals in the LI paradigm ignore the previously
irrelevant stimulus following pre-exposure [17].
In order to test the predictions arising from the Gray
et al. [20] model, human versions of these animal
behavioural paradigms were developed (e.g. Ref. [1]
for LI; Ref. [30] for KBE). The human tasks share the
logical structure of the animal tasks but vary greatly in
the actual behaviours asked of the participant. Three
key predictions arise from the model. First, that all three
tasks will be abolished in healthy volunteers by a low,
but not high, dose of oral amphetamine. The inverse
dose dependence was observed in the rat experiments
using LI [64]; whilst PREE is abolished by low doses of
amphetamine [13,63] under certain conditions [15] (see
Section 6). This hypothesis is as yet untested in KBE.
These effects are consistent with an action of amphetamine to release dopamine preferentially in nucleus
accumbens [64,65]. Second, and perhaps most fundamental to the model, is the prediction that the behavioural phenomena of LI, KBE and PREE would be
absent in acute schizophrenics, but restored in chronic
schizophrenics maintained on dopamine-blocking neuroleptic medication. Third, but more tentatively, is the
prediction that the tasks might be reduced in normal
subjects with high scores on personality traits associated
with vulnerability to psychosis (schizotypy).
For LI all three predictions have some support. LI
can be abolished by low, but not high, doses of
amphetamine [24,58]. Some studies have shown that
LI is abolished in acute, but not chronic, schizophrenics
[1,22,26,39,48], whereas others have failed to demonstrate this [57,67]. Finally, a number of studies using a
variety of different schizotypal measures have shown
that LI is abolished in people scoring high on these
measures [1,5,11,21,27,35,38], although a simple linear
relationship may not be present [4,68].
Human research has also shown that KBE is absent in
acute schizophrenia but is intact in chronic schizophrenia [3,30,31,43]. Against predictions, KBE was unaffected by high or low doses of amphetamine [23,31] and
is only marginally reduced in normal subjects scoring
high on measures of schizotypy [29].
To date there has been no attempt to test the third of
these key behavioural paradigms, the PREE. This is the
aim of the current series of studies. Like LI and KBE,
the PREE is reduced or abolished in animals by damage
to the septo-hippocampal system [18,28,53]. In addition,
it has been demonstrated [50] that the PREE is
abolished by section of the projection from the subiculum to nucleus accumbens, the pathway treated
within the Gray et al. [20] model as the most important
link between the limbic system and basal ganglia. This
effect has not been shown for either LI or KBE.
It can therefore be predicted from the Gray et al. [20]
model that the human PREE: (1) should be abolished by
a low, but not high, dose of oral amphetamine; (2)
should be absent in acute, but not chronic, schizophrenic patients; and (3) should be reduced in subjects high
on schizotypy. Amphetamine abolishes the PREE in the
rat by reducing resistance to extinction selectively in the
partial reinforcement (PR) condition. Therefore the
predictions for abolition of the PREE in acute schizophrenia, high schizotypal individuals, and amphetamine-treated healthy volunteers require that this takes
the form of a reduction of resistance to extinction
selectively in the PR group, but not in CR controls.
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
2. General method
Participants in all three studies were given a written
information sheet about the aims and nature of the
study and asked to provide written consent. Informed
consent was also obtained from patients’ responsible
medical officers for the schizophrenia study. For the
amphetamine study, participants were screened for
contraindications to amphetamine and for illegal drug
use (via both a clinical interview and urine test).
Repeated measures of heart rate and blood pressure
were taken throughout the amphetamine trial to check
for abnormal reactions to the drug. None occurred.
The method used was a modified version of one
developed by Vogel /Sprott [59]. To estimate the magnitude of the PREE we measured resistance to extinction after CR and PR training.
The task and instructions were presented by an Atari
1040ST microcomputer and colour monitor connected
to a four-button response box. There were two phases:
acquisition and extinction. The participant was told to
type in a sequence of four digits. The four buttons,
numbered 1 /4, were to be used, and the participant
could enter any order of the four digits, but could enter
each digit only once in any one sequence. The participant was further told that: the aim was to discover a
correct sequence, of which there might be more than
one; 20 pence would be paid for each correct sequence
entered; and that all money won would be received on
termination of the task. Throughout the task the total
amount accumulated up to that point was numerically
displayed at the top of the monitor, and further as a bar
graph along the right edge of the screen. The ‘correct’
sequence was defined as the fifth distinct sequence
entered by the participant. Correct responses were
followed by a ‘beep’, the word ‘correct’ on the monitor,
and incrementation of the cumulative displays. During
acquisition the CR group was rewarded in this way for
every correct response. The PR group was rewarded for
50% of the correct responses, on a quasi-random
schedule. When any sequence other than the correct
response was entered, and on the 50% of non-rewarded
target responses for the PR group, ‘incorrect’ was
displayed on the monitor. Extinction began immediately
after each participant reached a criterion of 20 target
responses. No further reward was given to participants
in either group regardless of the type of response. No
indication was given of the change from acquisition to
extinction. When the subject had entered 20 sequences
of any type during extinction, the experiment was
terminated. The word ‘jackpot’ appeared on the monitor, and the total amount of money won was incremented to £5.00 for each subject. The number of target (i.e.
previously rewarded) responses emitted during extinction was recorded, constituting the main measure of the
PREE.
335
In order to match our groups on possible confounding variables, measures of verbal intelligence (Mill Hill
Vocabulary Scale [49]) and age were taken.
3. Experiment 1 */the effects of low and high dose
amphetamine on the PREE
Both LI [54] and the KBE [9] are abolished by the
indirect dopamine agonist, amphetamine, in the rat. LI
is also abolished in normal human subjects by oral damphetamine [24,34,58]. As in the rat [64,66], the
abolition of human LI occurred at a relatively low
dose (5 mg) but not a high dose (10 mg) of amphetamine
[24]. Given the hypothesis that LI and the PREE depend
upon similar underlying neural processes [20], we therefore predicted that the PREE would similarly be
abolished by 5 but not 10 mg oral amphetamine.
3.1. Method
One hundred and twenty eight normal volunteers (65
men, 63 women), recruited by advertisement in a local
newspaper and paid £50.00, were randomly assigned to
three drug conditions under a double-blind protocol: 0,
5 and 10 mg oral d -amphetamine. Within each drug
condition, participants were assigned to CR and PR
conditions, matching as far as possible for age, and
verbal intelligence (Table 1). Exclusion criteria were a
history of mental illness, drug or alcohol dependency, or
marked abnormalities with hearing or vision. In addition, participants were excluded if they had a body
weight index outside the normal range (20 /28), because
of the effect of this upon amphetamine metabolism. All
participants were also screened for contra-indications to
amphetamine and for illegal drug use [24].
All participants also took part, in the same session, in
an experiment on LI or KBE. The data were initially
collected as two separate experiments. The first experiment investigated the PREE 45 min (N /60) after drug
administration, with the LI task being studied subsequently. The second experiment (n /68), measured the
Table 1
Means and standard errors (in parentheses) for age, verbal intelligence,
and plasma amphetamine levels scores across the experimental
conditions in experiment 1
0 mg
CR
PR
CR
PR
CR
PR
Age
Age
Verbal IQ
Verbal IQ
Plasma level
Plasma level
29.7
32.6
107.9
110.1
7.6
7.0
5 mg
(1.6)
(2.4)
(1.6)
(1.8)
(0.9)
(1.0)
31.0
28.1
110.1
106.4
17.3
14.4
10 mg
(1.1)
(1.2)
(2.0)
(1.4)
(2.0)
(2.2)
32.5
30.6
108.9
107.0
(1.9)
(1.1)
(1.7)
(1.7)
Amphetamine levels measured in mg/l. Verbal intelligence is measured by the Mill Hill Vocabulary Scale.
336
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
PREE 110 min after drug administration, immediately
following the KBE task. Venepuncture for amphetamine
plasma analysis (by gas chromatography; detection limit
1 /2 mg/l) was made immediately before the PREE task
began.
3.2. Results
Table 1 shows plasma amphetamine levels across
reinforcement schedule and drug group. The number
of trials to extinction is illustrated in Fig. 1. Due to gross
deviations in the variances and significant variation
from normal distributions, the data were analysed by
the methods recommended by Conover and Iman [8]:
the extinction scores were ranked and an analysis of
variance then performed. A three-way ANOVA was
performed first with reinforcement schedule (CR or
PR), amphetamine dose (0, 5 or 10 mg) and experiment
(delay of 45 or 110 min) as variables. The variable of
experiment had no main effect nor did it interact with
any other variables, hence data were collapsed across
this variable and a two-way ANOVA was performed. As
expected there was a robust effect of reinforcement
schedule, with slower extinction for the PR group
[F (1, 122) /163.39, P B/0.0001] thus replicating the
well documented PREE. However there was no effect
of amphetamine dose [F (2, 122) B/1], nor any interaction between reinforcement schedule and drug dose
[F (2, 122) /2.03, ns].
3.3. Discussion
The magnitude of the PREE was unchanged by either
5 or 10 mg d-amphetamine. Since the 5 mg treatment
Fig. 1. Mean number of target responses during extinction as a
function of the dose of d -amphetamine in experiment 1. Error bars
indicate 91 standard error of the mean. CR, continuous reinforcement; PR, partial reinforcement.
modified LI in the same participants, tested in the same
session [24], this negative result cannot be due to the use
of a generally ineffective pharmacological manipulation.
The same protocol was also used to evaluate the effects
of low and high dose amphetamine on the KBE and, as
with the present study on the PREE, KBE was
unaffected by either 5 or 10 mg d -amphetamine [23].
The lack of effect of amphetamine on the PREE is
inconsistent with the hypothesis that the PREE and LI
reflect similar neural processes.
4. Experiment 2 */the effects of acute and chronic
schizophrenia on the PREE
Both LI [1,22] and the KBE [30] are absent in acute
schizophrenic patients, but present in chronic schizophrenic patients maintained on neuroleptic medication.
If the PREE is mediated by the same neural processes as
these phenomena then the PREE should also be absent
in acute, but not chronic, schizophrenic patients. This
prediction was evaluated in Experiment 2. A group of
healthy control participants was also tested on the
PREE to see if a diagnosis of schizophrenia per se had
any effect on the PREE.
4.1. Method
Participants were 16 acute (12 male, 4 female) and 16
chronic (12 male, 4 female) schizophrenic in-patients
who met Research Diagnostic Criteria [55]. The acute
schizophrenic patients were either suffering from their
first psychotic breakdown (N /9) or were in an acute
phase of an otherwise chronic disorder (N /7). All
patients were tested in the first 2 weeks following the
start of their current in-patient admission and commencement of neuroleptic medication. The chronic
schizophrenic patients had been continuously ill for at
least 6 months and had received neuroleptic medication
for at least this period of time. All the schizophrenic
patients were severely ill, with high levels of both
positive and negative symptoms. Considerable care
was taken to match patients across groups in terms of
the presence and severity of positive and negative
symptoms in order to try to avoid the potential
confound of significant differences in degree of symptomatology. Psychotic symptoms were assessed by the
Brief Psychiatric Rating Scale [45], scored separately for
positive and negative symptoms [44], and by the
MAINE scale of paranoid and non-paranoid schizophrenia [40]. Most of the schizophrenic patients were
receiving phenothiazine drugs, but two of the acute
patients were unmedicated and one had never received
neuroleptic medication. Table 2 depicts the demographic and clinical information for the schizophrenic
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
337
Table 2
Means and standard errors (in parentheses) for the demographic and clinical data for schizophrenic participants across experimental condition in
experiment 2
Age
Verbal IQ
Medication
Age 1st Hosp.
Positive
Negative
Paranoid
Non-paranoid
Acute schizophrenia
Chronic schizophrenia
Healthy controls
CR
PR
CR
PR
CR
PR
31.6 (3.9)
93.0 (4.1)
612 (175)
22.9 (0.9)
48.9 (2.5)
24.6 (3.3)
19.6 (1.5)
16.8 (1.2)
34.3 (6.7)
94.0 (4.6)
450 (121)
31.1 (7.1)
36.8 (1.5)
15.8 (1.6)
13.9 (1.3)
9.6 (0.9)
31.6 (3.3)
89.0 (3.4)
700 (66)
21.4 (1.3)
32.4 (3.2)
23.1 (2.2)
13.3 (1.8)
12.9 (1.7)
34.4 (3.4)
87.0 (3.2)
600 (115)
24.1 (2.8)
36.6 (3.2)
23.8 (3.4)
17.0 (2.2)
15.8 (1.6)
24.4 (2.4)
99.0 (3.3)
26.2 (2.7)
100.0 (2.6)
CR, continuous reinforcement; PR, partial reinforcement. Verbal intelligence as measured by the Mill Hill Vocabulary Scale. Dose of
phenothiazine medication converted to chlorpromazine equivalents in mg following [10]. Positive and negative symptoms assessed on the Brief
Psychiatric Rating Scale. Paranoid and non-paranoid symptoms assessed by the MAINE scale.
patient groups. In addition we also tested 64 (32 male,
32 female) healthy control participants*/see Table 2.
4.2. Procedure
The PREE was tested as in the experiments reported
above; all participants were also tested on LI (immediately after the PREE), as reported by in Gray et al. [22].
The conditions in both the PREE and LI experiments
were counterbalanced, such that half the sample completed the PREE experiment first and half the LI
experiment first. Likewise, half the people who received
CR in the PREE task were also in the pre-exposed
condition of the LI task, whereas the other half were in
the non-preexposed condition. Analysis of extinction
scores revealed no significant effects of task order and
therefore this variable is ignored here.
4.3. Results
The mean responses in extinction for the three groups
after the conditions of continuous and partial reinforcement are illustrated in Fig. 2. Once again the data
showed gross inhomogeneity of variance between conditions, and significant deviations from normality.
Hence the methods of Conover and Iman [8] were
applied.
As expected we obtained far greater scores for the PR
group in comparison to the CR group [F (1, 90) /34.27,
P B/0.001], demonstrating a robust PREE. In addition
we also found an effect of diagnosis [F (2, 90) /3.64,
P B/0.05] but no interaction between reinforcement
schedule and diagnosis [F (2, 90) /1.83, ns]. Examination of Fig. 2 suggests the effect of diagnosis is due to
greater resistance to extinction for the schizophrenic
patients compared to the controls, particularly for the
CR group (though we note the lack of significant
interaction). This was supported by Tukey HSD tests
Fig. 2. Mean number of target responses during extinction as a
function of the diagnosis of the participants in experiment 2. Error
bars indicate 91 standard error of the mean. CR, continuous
reinforcement; PR, partial reinforcement.
that showed that in the CR condition the acute
schizophrenic patients had greater scores than the
healthy controls (P B/0.05), whilst the chronic schizophrenic patients had marginally greater scores than the
healthy controls (P B/0.09). There was no significant
difference in the CR condition between the two groups
of schizophrenic patients. No differences were observed
for the PR condition across any of the groups.
4.4. Discussion
Experiment 2 failed to find a significant reduction in
PREE in acute as compared to chronic schizophrenic
patients. There was some hint that the PREE was
smaller for schizophrenic patients per se. However,
analysis showed that even this marginal effect was due
338
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
to increased resistance to extinction in the CR schedule
rather than reduced resistance to extinction in the PR
schedule as predicted.
The acute and chronic patients were statistically
indistinguishable from one another, whereas in our
parallel experiments using the LI [1,22] and KBE [30]
paradigms, only acute schizophrenic patients differed
from controls, showing an abolition of these behavioural
phenomena. This discrepancy cannot reflect differences
in the patient groups studied, since one of the relevant
LI experiments [22] used the same participants as those
studied here and, indeed, LI and the PREE were
measured in the same session.
5. Experiment 3 */PREE and the effect of schizotypy
On the assumption that psychotic tendencies exist on
a continuum [7] it is predicted that those scoring highly
on such measures will show comparable patterns of
behaviour to those of schizophrenic patients. Thus a
number of studies have demonstrated that LI is reduced
in such individuals [1,11,21,35,38], as is KBE [29]. In this
experiment we test the prediction that PREE will also be
reduced in individuals scoring highly on a measure of
schizotypy.
5.1. Method
Participants were 64 paid normal volunteers (32 male,
32 female) obtained from an employment agency.
Exclusion criteria were a history of mental illness,
drug or alcohol dependency, or marked abnormalities
with hearing or vision. As far as possible participants
were matched across experimental conditions (CR and
PR training) for sex, age, and verbal intelligence as
tested by the Mill Hill Vocabulary Scale [49]. At the end
of behavioural testing participants completed the Eysenck Personality Questionnaire (EPQ; [12]) in order to
assess Psychoticism (P) scores. Whilst the P-scale of the
EPQ is somewhat controversial as a measure of schizotypy (see [21] for discussion), these experiments were
performed before later conceptualisations of schizotypy
as a multi-dimensional construct [7], and the P-scale had
already been shown to be related to LI [2] and the KBE
[29]. Given the success of the P-scale in predicting
reduced LI and KBE, it is the logical first candidate
for examination of the PREE and schizotypy. Table 3
presents the means and standard errors of the age,
verbal IQ and P-scale measures across experimental
condition.
Participants also took part in an experiment on LI,
using the procedures described by [24]. Order of testing
on the LI and PREE paradigms was randomly counterbalanced; no significant order effects were observed and
this variable is therefore ignored here.
Table 3
Means and standard errors (in parentheses) for demographic and
personality characteristics across the experimental conditions in
experiment 1
Age
Verbal IQ
P-scale
CR
PR
24.8 (1.2)
103.7 (2.2)
4.6 (0.9)
24.7 (1.3)
107.2 (2.0)
4.5 (0.8)
Verbal intelligence was measured by the Mill Hill Vocabulary Scale.
5.2. Results
Participants were assigned to either a ‘low’ or ‘high’
schizotypy group according to a median split based
upon their P-score. Those participants (n/9) with
exactly the median P-score were eliminated from the
analysis.
The mean responses in extinction for the two groups
after the conditions of continuous and partial reinforcement are illustrated in Fig. 3. Once again the data had
gross inhomogeneity of variance between conditions,
and significant deviations from normality. Hence the
methods of Conover and Iman [8] were applied.
Once more we obtained a robust PREE [F (1, 51) /
57.62, P B/0.0001]. Overall the low and high schizotypy
groups had similar mean target responses [F(1, 51) B/1].
Crucially there was a significant interaction between
schizotypy group and reinforcement schedule
[F (1, 51) /4.13, P B/0.05]. Examination of Fig. 3 suggests a smaller PREE for the low compared to the high
schizotypy group, and that the effect is due to a decrease
in score with increasing schizotypy for the CR group,
but an increase for the PR group. Examination of this
interaction showed that those with high scores on the P-
Fig. 3. Mean number of target responses during extinction as a
function of the median split of P-scores in experiment 3. Error bars
indicate 91 standard error of the mean. CR, continuous reinforcement; PR, partial reinforcement.
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
scale had a smaller learning score than the low scorers in
the CR condition [t(24) /2.33, P B/0.05] whilst the
groups did not differ in the PR condition [t (27) /0.46,
ns]. This pattern of results was confirmed by correlating
the P-score with the extinction score for the PR and CR
groups separately. For the CR group, Spearman’s Rho
indicated a significant negative relationship [Rho //
0.38, n/32, P B/0.05] whilst no significant correlation
existed in the PR group [Rho /0.15, n/32, ns].
5.3. Discussion
Our findings with regard to schizotypal personality
run counter to our predictions. To the extent that we
observed any effect of schizotypy, this was in the
direction opposite to observations made with the human
LI [2,38] and KBE [29] paradigms. Both the latter two
phenomena are reduced in participants high on the
Psychoticism scale of the EPQ, whereas in the present
experiment the PREE was greater in such participants.
Moreover the difference was due to changes in extinction score for the CR group (in which resistance to
extinction decreased with increasing P-score) rather than
in the PR group as predicted.
6. General discussion
In a series of three experiments examining the human
PREE we have found that:
1) low and medium doses of amphetamine have no
effect on resistance to extinction in either the CR or
PR condition;
2) both chronic and acute schizophrenics show a
reduced PREE due to greater resistance to extinction in the CR condition compared to controls; and
3) those high on schizotypy show a greater PREE due
to less resistance to extinction in the CR condition.
The experiments were performed to test predictions
arising from Gray et al.’s [20] theory of the neuropsychology of schizophrenia. This theory predicted that the
PREE would be reduced (1) by low, but not high, doses
of amphetamine, (2) in acute but not chronic schizophrenia, and (3) in individuals scoring high on a
schizotypy scale. Moreover all these reductions should
be due to reduced resistance to extinction in the PR
group, with no change in the CR group. It is clear that
the results provide no support for this model.
There seem to be three possible reasons for this
outcome. The first is that the theory of Gray et al. [20]
is, at least in part, wrong.
Secondly, the PREE task may not be measuring the
same psychological process as the LI and KBE tasks.
339
The latter effects occur when stimuli lose the capacity to
enter into associations either because initially they
appear to have no consequence (LI), or no predictive
value (KBE). These processes explain why the stimulus
subsequently fails to gain attention when it later
becomes relevant. Thus, behaviour is absent because
attentional resources are focussed away from the
‘‘thought-to-be-irrelevant’’ stimuli. In contrast, in the
PREE, the PR subject is able to continue responding
regardless of the occurrence of non-reward (this having
lost salience due its having been previously interspersed
with reward). Thus, behaviour is present because attentional resources are focussed away from the ‘‘thoughtto-be-irrelevant’’ consequence of that behaviour.
Thirdly, though we purport to be measuring the
PREE in human subjects, it is far from clear that the
human PREE is the same as the PREE measured in
other animals (mainly rats). Whilst many aspects of the
findings of PREE in human and rat are similar
[16,42,52], there are also striking differences between
the findings [16,46,47]. The task we used was one well
established in the literature [59]. However, this does not
necessarily mean that it measures the PREE. One
possible confounding factor in our task [59] is that the
target response is exactly the same each time. It is well
documented that many schizophrenic patients exhibit
the phenomenon of perseveration, thought to reflect
hypodopaminergic activity in the frontal cortex [60,61].
Hence schizophrenic patients may well tend to keep
producing the learned response even in the absence of
reward due to a process of perseveration rather than due
to resistance to extinction. This may go some way to
explaining why we found greater resistance to extinction
in the CR schedule for the schizophrenic patients.
With regard to the amphetamine experiment, it is also
the case that there are many different conditions under
which amphetamine abolishes the PREE in the rat. In
the current study we have only investigated two of these
conditions (a single low versus high dose of amphetamine administered in both acquisition and extinction).
The amphetamine PREE studies in the rat present a
complex picture of the conditions under which the
PREE is attenuated following amphetamine. For example, Weiner et al. [62] found that repeated low doses
of amphetamine administered in the acquisition phase of
the task abolished the PREE, irrespective of drug
treatment in extinction. Conversely, a PREE was
obtained in animals that received saline in acquisition,
independent of drug treatment in extinction. Amphetamine administered in extinction alone was found to
increase resistance to extinction in PR animals while
having no effect on CR animals. On the basis of these
results, Weiner et al. [62] proposed that amphetamine
disrupts performance in conflict situations that involve
competing contingencies of reinforcement and non-
340
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
reinforcement and where the appropriate performance
requires the subject’s responding to come under the
control of the non-reinforcement contingency. Consistent with this analysis, Weiner et al. [63] found in a one
trial a day paradigm that amphetamine reduces control
over behaviour by stimuli associated with non-reinforcement without affecting the capacity of reinforcement
to control behaviour. Amphetamine was administered
to PR animals only on reinforced or only on nonreinforced trials. PR animals that were given placebo on
non-reinforced trials showed a strong PREE whereas
those who received amphetamine on non-reinforced
trials showed no PREE, irrespective of the drug injected
on reinforced trials. Again, there was found to be no
effect of amphetamine on the CR animals. The neuroleptic drug haloperidol was also found to have an action
on non-reinforcement and not reinforcement [14]. Thus,
haloperidol increased the magnitude of the PREE by
increasing the rate of extinction and this effect was
entirely due to the administration of the drug in
extinction, independent of drug treatment in acquisition.
This complex analysis of the conditions under which
amphetamine abolishes the PREE in the rat contrasts
with our current study where we have only investigated
the effects of amphetamine on the PREE under one set
of conditions (e.g. a single low vs. high dose of
amphetamine administered in both acquisition and
extinction). It is, however, difficult to imagine how a
‘‘one trial a day’’ human paradigm could be developed
and, of course, chronic amphetamine administration in
humans would be impossible due to ethical considerations.
Given the importance of the PREE to theories of
schizophrenia, and the problems we have identified
above, it may be unwise to discard it too soon. New
PREE tasks are being developed that seem free from
previous shortcomings, and may allow both within- and
between-subject tasks (e.g. [56]). Such a within-subject
task would be a valuable tool for researchers, but has
proved difficult to establish using the LI [25,36] or KBE
[31] paradigms.
A final noteworthy feature of our results is that, as is
the case also for LI [51], the PREE task disclosed
differences in the behaviour of schizophrenic patients
(increased resistance to extinction, especially in the CR
condition; Fig. 2) and normal individuals identified by
questionnaire as schizotypal (reduced resistance to
extinction in the CR condition; Fig. 3), in each case
relative to the behaviour of controls. This divergence is
inconsistent with proposals that see schizophrenia as an
extreme form of the same underlying behavioural trait
that constitutes schizotypy [6], although the effect of
neuroleptic medication in schizophrenic patients on
these tasks also needs to be taken into account.
Acknowledgements
Thanks are due to Professor J.D. Parkes for advice in
the design of the psychopharmacological studies. We are
grateful to the Medical Research Council and to
Bristol /Myers Squibb Pharmaceutical Corporation for
financial support and to SmithKline Beecham plc for
donating amphetamine.
References
[1] Baruch I, Hemsley DR, Gray JA. Differential performance of
acute and chronic schizophrenics in a latent inhibition task.
Journal of Nervous and Mental Disease 1988;176:598 /605.
[2] Baruch I, Hemsley DR, Gray JA. Latent inhibition and ‘psychotic
proneness’ in normal subjects. Personality and Individual Differences 1988;9:777 /83.
[3] Bender S, Muller B, Oades RD, Sartory G. Conditioned blocking
and schizophrenia: a replication and study of the role of
symptoms, age, onset-age of psychosis and illness-duration.
Schizophrenia Research 2001;49:157 /70.
[4] Braunstein-Bercovitz H. Is the attentional dysfunction in schizotypy related to anxiety? Schizophrenia Research 2000;46:255 /67.
[5] Braunstein-Bercovitz H, Lubow RE. Are high schizotypal-normal
participants distractible or limited in attentional resources? A
study of Latent Inhibition as a function of masking task load and
schizotypy level. Journal of Abnormal Psychology 1998;107:659 /
70.
[6] Claridge G. Theortical background and issues. In: Claridge G,
editor. Schizotypy. Implications for Illness and Health, 1997:377 /
86.
[7] Claridge GS, McCreery C, Mason O, Bentall RP, Boyle G, Slade
PD, Popplewell D. The multidimensional nature of schizotypal
traits: a factor analytic study with normal subjects. British Journal
of Clinical Psychology 1996;35:103 /15.
[8] Conover WJ, Iman RL. Ranked transformations as a bridge
between parametric and non-parametric statistics. American
Statistician 1981;35:1240 /9.
[9] Crider A, Solomon PR, McHahon MA. Disruption of selective
attention in the rat following d -amphetamine administration:
relationship to schizophrenic attention disorder. Biological Psychiatry 1982;7:351 /61.
[10] Davis JM. Comparative doses and costs of antipsychotic medication. Archives of General Psychiatry 1976;33:858 /61.
[11] De la Casa GL, Ruiz G, Lubow RE. Latent inhibition and recall/
recognition of irrelevant stimuli as a function of preexposure
duration in high and low psychotic prone normal subjects. British
Journal of Psychology 1993;84:119 /32.
[12] Eysenck HJ, Eysenck SBG. Manual of the Eysenck Personality
Questionnaire. London: Hodder and Stoughton, 1975.
[13] Feldon J, Bercovitz H, Weiner I. The effects of amphetamine on a
multi-trial partial reinforcement extinction effect (PREE) in a
runway. Pharmacology. Biochemistry and Behaviour 1989;32:65 /
9.
[14] Feldon J, Weiner I. Effects of haloperidol on a multitrial partial
reinforcement extinction effect (PREE)-evidence for neuroleptic
drug-action on nonreinforcement but not on reinforcement.
Psychopharmacology 1991;105:407 /14.
[15] Feldon J, Weiner I. Amphetamine and the multitrial partialreinforcement extinction effect (PREE) in an operant chamberprocedural modifications that lead to an attenuation of the PREE.
Pharmacology Biochemistry and Behaviour 1992;41:309 /15.
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
[16] Flora SR, Pavlik WB. Conventional and reversed partial reinforcement effects in human operant responding. Bulletin of the
Psychonomic Society 1990;28:429 /32.
[17] Gray JA. Elements of a two-process theory of learning. London:
Academic Press, 1975.
[18] Gray JA. The neuropsychology of anxiety: an enquiry into the
functions of the septohippocampal system. Oxford, England:
Oxford University Press, 1982.
[19] Gray JA. Integrating schizophrenia. Schizophrenia Bulletin
1998;24:249 /66.
[20] Gray JA, Feldon J, Rawlins JNP, Hemsley DR, Smith AD. The
neuropsychology of schizophrenia. Behavioral and Brain Sciences
1991;14:1 /20.
[21] Gray NS, Fernandez M, Williams J, Ruddle RA, Snowden RJ.
What schizotypal dimensions abolish latent inhibition? British
Journal of Clinical Psychology, in press.
[22] Gray NS, Hemsley DR, Gray JA. Abolition of latent inhibition in
acute, but not chronic, schizophrenics. Neurology and Psychiatric
Brain Research 1992;1:83 /9.
[23] Gray NS, Pickering AD, Gray JA, Jones SH, Abrahams S,
Hemsley DR. Kamin blocking is not disrupted by amphetamine in human subjects. Journal of Psychopharmacology
1997;11:301 /11.
[24] Gray NS, Pickering AD, Hemsley DR, Dawling S, Gray JA.
Abolition of latent inhibition by a single 5 mg dose of d amphetamine. Psychopharmacology 1992;107:425 /30.
[25] Gray NS, Pilowsky LS, Gray JA, Kerwin RW. Latent inhibition
in drug naive schizophrenics: relationship to duration of illness
and dopamine D2 binding using SPET. Schizophrenia Research
1995;17:95 /107.
[26] Guterman Y, Josiassen RC, Bashoire TE, Johnson M, Lubow
RE. Latent inhibition effects reflected in event-related brain
potentials in healthy controls and schizophrenics. Schizophrenia
Research 1996;20:315 /26.
[27] Hofer I, Della Casa V, Feldon J. The interaction between
schizotypy and latent inhibition: modulation by experimental
parameters.
Personality
and
Individual
Differences
1999;26:1075 /88.
[28] Jarrard LE, Feldon J, Rawlins JNP, Sinden JD, Gray JA. The
effects of intrahippocampal ibotenate on resistance to extinction
after continuous or partial reinforcement. Experimental Brain
Research 1986;61:519 /30.
[29] Jones SH, Gray JA, Hemsley DR. The Kamin blocking effect,
incidental learning and schizotypy. Personality and Individual
Differences 1992;13:57 /60.
[30] Jones SH, Gray JA, Hemsley DR. Loss of Kamin blocking effect
in acute but not chronic schizophrenics. Biological Psychiatry
1992;32:739 /55.
[31] Jones SH, Hemsley DR, Ball S, Serra A. Disruption of the Kamin
blocking effect in schizophrenia and in normal subjects following
amphetamine. Behavioural Brain Research 1997;88:103 /14.
[32] Kamin LJ. ’Attention-like’ processes in classical conditioning. In:
Jones MR, editor. Miami symposium on the prediction of
behaviour: aversive stimulation. Miami, FL: University of Miami
Press, 1968.
[33] Kelly AE, Domesick VB. The distribution of the projection from
the hippocampal formation to the nucleus accumbens in the rat.
An anterograde- and retrograde-horseradish peroxidase study.
Neuroscience 1982:2321 /35.
[34] Kumari V, Cotter PA, Mulligan OF, Checkley SA, Gray NS,
Hemsley D, Thornton JC, Corr PJ, Toone BK, Gray JA. Effects
of d -amphetamine and haloperidol on latent inhibition in healthy
male volunteers. Journal of Psychopharmacology 1999;13:398 /
405.
[35] Lipp OV, Vaitl D. Latent inhibition in human Pavlovian
differential conditioning: effect of additional stimulation after
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
341
pre-exposure and relation to schizotypal traits. Personality and
Individual Differences 1992;13:1003 /12.
Lubow RE. Latent inhibition as a measure of learned inattention:
some problems and solutions. Behavioural Brain Research
1997;88:75 /83.
Lubow RE, Gerwitz JC. Latent inhibition in humans: data,
theory, and implications for schizophrenia. Psychological Bulletin
1995;117:87 /103.
Lubow RE, Ingberg-Sachs Y, Zalstein-Orda N, Gewirtz JC.
Latent inhibition in low and high ‘psychotic prone’ normal
subjects. Personality and Individual Differences 1992;13:563 /72.
Lubow RE, Kaplan O, Abramovich P, Rudnick A, Laor N.
Visual search in schizophrenia: latent inhibition and novel popout effects. Schizophrenia Research 2000;45:145 /56.
Magaro PA, Abrams L, Cantrell P. The MAINE scale
of paranoid and non-paranoid schizophrenia: reliability and
validity. Journal of Consulting and Clinical Psychology
1981;49:439 /47.
Mogenson GJ, Wang CR, Yim CY. Influence of dopamine on
limbic inputs to the nucleus accumbens. Annals of the New York
Academy of Science 1988;537:86 /100.
Nevin JA. The momentum of compliance. Journal of Applied
Behavior Analysis 1996;29:535 /47.
Oades RD, Zimmermann B, Eggers C. Conditioned blocking in
patients with paranoid, non-paranoid psychosis or obsessive
compulsive disorder: Associations with symptoms, personality
and monoamine metabolism. Journal of Psychiatric Research
1996;30:369 /90.
Overall JE. The Brief Psychiatric Rating Scale in psychopharmacological research. In: Pichot P, Oliver-Martin R, editors.
Psychological measurements in psychopharmacology. Modern
problems in pharmacopsychiatry, vol. 7. Basel: Karger,
1974:67 /78.
Overall JE, Gorham DR. The Brief Psychiatric Rating Scale.
Psychological Reports 1962;10:799 /812.
Pavlik WB, Flora SR. Human responding on multiple variable
interval schedules and extinction. Learning and Motivation
1993;24:88 /99.
Pittenger DJ, Pavlik WB. Analysis of the partial reinforcement
extinction effect in humans using absolute and relative comparisons of schedules. American Journal of Psychology 1993;101:1 /
14.
Rascle C, Mazas O, Vaiva G, Tournant M, Raybois O,
Goudemand M, Thomas P. Clinical features of latent inhibition
in schizophrenia. Schizophrenia Research 2001;51:149 /61.
Raven JG. Manual for Raven’s Progressive Matrices and
vocabulary scales. London: HK Lewis, 1981.
Rawlins JNP, Feldon J, Tonkiss J, Coffey PJ. The role of
subicular output in the development of the partial reinforcement
extinction effect. Experimental Brain Research 1989;77:153 /60.
Serra AM, Jones SH, Toone B, Gray JA. Impaired associative
learning in chronic schizophrenics and their first-degree relatives:
a study of latent inhibition and the Kamin blocking effect.
Schizophrenia Research 2001;28:273 /89.
Shigley R, Guffey J. Resistance to extinction as a function of
partial reinforcement before and after continuous reinforcement
in a human operant task. Journal of General Psychology
1978;99:257 /61.
Sinden JD, Jarrard LE, Gray JA. The effects of intra-subicular
ibotenate on resistance to extinction after continuous or partial
reinforcement. Experimental Brain Research 1988;73:315 /9.
Solomon PR, Crider A, Winkelman JW, Turi A, Kamer RM,
Kaplan LJ. Disrupted latent inhibition in the rat with chronic
amphetamine or haloperidol- induced supersensitivity: relationship to schizophrenic attention disorder. Biological Psychiatry
1981;16:519 /37.
342
N.S. Gray et al. / Behavioural Brain Research 133 (2002) 333 /342
[55] Spitzer RL, Endicott J, Robbins E. Research Diagnostic Criteria
(RDC) for a selected group of functional disorders, 3rd ed.. New
York: State Psychiatric Institute, 1978.
[56] Svartdal F. Persistence during extinction: conventional and
reversed PREE under multiple schedules. Learning and Motivation 2000;31:21 /40.
[57] Swerdlow NR, Braff DL, Hartston H, Perry W, Geyer MA.
Latent inhibition in schizophrenia. Schizophrenia Research
1996;20:91 /103.
[58] Thornton JC, Dawe S, Lee C, Capstick C, Porr PJ, Cotter P,
Frangou S, Gray NS, Russell MAH, Gray JA. Effects of nicotine
and amphetamine on latent inhibition in human subjects.
Psychopharmacology 1996;127:164 /73.
[59] Vogel-Sprott MD. Partial-reward training for resistance to
punishment and to subsequent extinction. Journal of Experimental Psychology 1967;75:13 /140.
[60] Weinberger DR. Schizophrenia and the frontal lobe. Trends in
Neurosciences 1988;11:367 /70.
[61] Weinberger DR, Berman KF. Prefrontal function in schizophrenia: confounds and controversies. Philosophical Transactions of
the Royal Society of London B 1996;351:1495 /503.
[62] Weiner I, Bercovitz H, Lubow RE, Feldon J. The abolition of the
partial reinforcement extinction effect (PREE) by amphetamine.
Psychopharmacology 1985;86:318 /23.
[63] Weiner I, Feldon J, Bercovitz H. The abolition of the partial
reinforcement extinction effect (PREE) by amphetamine: disruption of control by non-reinforcement. Pharmacology Biochemistry and Behaviour 1987;27:205 /10.
[64] Weiner I, Feldon J, Izraeli-Telerant A. Latent inhibition is not
affected by acute or chronic administration of 6mg/kg d amphetamine. Psychopharmacology 1987;91:345 /51.
[65] Weiner I, Lubow RE, Feldon J. Chronic amphetamine and latent
inhibition. Behavioural Brain Research 1981;2:285 /6.
[66] Weiner I, Lubow RE, Feldon J. Disruption of latent inhibition by
acute administration of low doses of amphetamine. Pharmacology
Biochemistry and Behaviour 1988;30:871 /8.
[67] Williams JH, Wellman NA, Geaney DP, Cowen PJ, Feldon J,
Rawlins JNP. Reduced latent inhibition in people with schizophrenia: an effect of psychosis or of its treatment. British Journal
of Psychiatry 1998;172:243 /9.
[68] Wuthrich V, Bates TC. Schizotypy and latent inhibition: nonlinear linkage between psychometric and cognitive markers.
Personality and Individual Differences 2001;30:783 /98.