In 1980, with the publication of the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) [1], the problem of excessive gambling was officially recognized as a psychiatric diagnosis. In DSM-IV-R [2] pathological gambling (PG) is defined as an impulse control disorder, characterized by persistent and maladaptive gambling behaviours, with disruptive consequences for familial, occupational and social functions. Epidemiological studies suggest that men, adolescents, ethnic minorities and patients with other psychiatric disorders are at a higher risk of suffering from PG [3], although more recent epidemiological data indicate that the disorder is progressing rapidly also amongst women and teen-agers [4, 5]. Other risk factors include early exposure to gambling opportunities, such as those offered by the internet, cognitive deficits and relatives with PG or alcohol dependence [6–9].

Several studies have documented that PG patients are often affected by other psychiatric disorders, in particular depression, alcoholism or drug abuse (DA) [10–13]; the comorbid disorders are explained either as secondary to PG or primarily responsible for PG onset. Furthermore, PG is interpreted as belonging to the group of impulsive, compulsive, addictive behaviours which may share common genetic and/or etiological mechanisms [14].

Although the pathophysiology of PG is unclear, probably it represents the result of the interplay between individual and environmental factors. Different hypotheses have been put forward: a genetic vulnerability mainly involving dopamine receptors, biochemical dysfunctions at the level of the serotonin and dopamine systems and/or alterations of different brain areas.

Patients with ventromedial prefrontal cortex abnormalities show a peculiar impairment in decision making, as assessed by the gambling task, a specific test that explores the ability to balance immediate rewards against long-term consequences [15–17]. PG patients, as well as patients suffering from obsessive-compulsive disorder (OCD) or cocaine, opiate or alcohol abuse, present similar alterations when presented with the gambling task: these subjects continue to make choices according to an immediate reward, despite being fully aware of the long-term negative consequences [18, 19]: taken together, these findings would suggest that all these heterogeneous patients may share a common dysfunction at the level of the ventromedial prefrontal cortex. In line with this hypothesis, a recent study using a comprehensive neuropsychological battery measuring executive functions, demonstrated that PG and alcohol-dependent patients showed a reduction of executive functioning performance on inhibition, time estimation, cognitive flexibility and planning tasks [20]. Moreover, neurocognitive indicators of decision-making and disinhibition, such as the Card Playing Task and Stop Signal Reaction Time respectively, seem to be powerful predictors of relapse in PG [21]. The impairement of decision-making observed in PG might be explained by the inability to inhibit irrelevant information: in a recent study, the performances on the reverse Stroop task, which highly discriminates the ability to inhibit interferences, were significantly impaired in PG patients than in healthy subjects [22].

PG has been associated to impulsivity and attention deficit: PG patients were found to perform significantly worse than control subjects on attention measures and showed more childhood behaviors related to attention deficits [23]. More recently, neuropsychological measures of impulsivity, such as the reaction time and number of errors at Go/No-go tasks, as well as the scores at the Barratt Impulsiveness Scale, were higher in PG patients than healthy control subjects [24], while highlighting the importance of this dimension in the clinical picture of PG.

Given the paucity of information in this field, our study aimed to evaluate the possible involvement of some brain areas in PG by means of a battery of neuropsychological tests, in particular the Wisconsin Card Sorting Test (WCST) [25], the Wechsler Memory Scale revised (WMS-R) [26] and the Verbal Associative Fluency Test (FAS) [27, 28].

Twenty outpatients (15 men, 5 women, mean age: 26 ± 4 years) with a diagnosis of PG according to the DSM-IVR criteria [2] were recruited at the Dipartimento di Psichiatria, Neurobiologia, Farmacologia e Biotecnologie at Pisa University, in the Psychiatry Section. None suffered from any severe physical illness nor had ever taken psychotropic drugs, except for two patients who had occasionally taken benzodiazepine for difficulty with sleeping. The mean score of PG at the Yale Brown for Obsessive Compulsive Disorder modified for PG (Y-BOCS-PG) [29] was 30 ± 2. The age of onset of the disorder (mean ± SD) was 17 ± 2 years.

Seven patients were suffering also from bipolar disorder (BD) of type II, 5 from OCD, 5 from DA (cocaine and cannabis) and 3 from alcohol abuse. Five patients had two comorbid disorders: 4 BD patients were suffering also from DA and 1 OCD patients from alcohol dependence.

The patients were compared with 20 matched healthy subjects (15 men, 5 women, mean age: 25 ± 5), who were selected from amongst a pool of 500 control individuals undergoing the same battery of neuropsychological tests. Both patients and control subjects were all right-handed.

The main bias of our study is that it was carried out in a small sample of PG patients: this was a consequence of the setting in which the patients were selected, namely a public psychiatric department, to which patients with such a disorder are generally sent on account of severe legal problems and/or as a result of strong prompting by their relatives. This might also provide an explanation for the high level of comorbid disorders observed; also, because of the small sample size, we were unable to determine the possible effect of different diagnostic patterns on neuropsychological tests. Nevertheless, our findings may be considered intriguing: in fact, our group of patients affected by PG undergoing a battery of neurological tests, namely the WCST, the WMS-R and the FAS, showed that they had sufficient or normal intellectual, linguistic and visual-spatial abilities. As far as the WCST is concerned, PG patients showed qualitative but not quantitative deficits: in fact, although no differences were found between PG patients and healthy control subjects in the total number of categories completed, different abnormalities were detected at some subscales. As compared with healthy subjects, the thinking of PG patients appeared perseverant, because when they tried to resolve a problem while using an incorrect method, they tended to continue beyond that point at which other subjects would have looked for alternative solutions. A similar behavior has been observed in PG patients at both the card-choosing tests [21] and the GO/NO-GO task [24]. The difficulty that PG patients showed in learning from their mistakes and in re-directing themselves in the appropriate direction represents one of the most characteristic features of patients with alterations of the prefrontal lobe. This aspect has been observed in a significant number of experimental paradigms, in particular, patients with lesions of the prefrontal lobe are sometimes able to identify correct answers, while nevertheless still continuing to produce wrong answers [31–36]. Given the possibility of disproportionate aging of the ventromedial prefrontal areas, which would implicate the decline of decision making with age [37], the sample we selected was homogeneous with regard to that variable.

Our findings are also compatible with other studies reporting worse performances in cognitive "risk-taking" tasks in patients with prefrontal lesions, as compared with healthy control subjects or patients with temporal lobe excision [38]. A link has also been underlined between attention problems or impulsivity in executive functions and minimal brain damage including impairment of the prefrontal lobe [39]; in some cases, the difficulties in controlling behaviour leads such patients to neglect or break the rules of the game [40, 41]. Previous studies, although reporting no alterations at the WMS-R and FAS similar to what described by us, could not detect dysfunctions at the WCST, but only at the gambling task in disagreement with our findings [42]. Since the WCST is sensitive to damage to the dorsolateral portion of the prefrontal cortex, as well as to damage to non-prefrontal cortical regions connected to the prefrontal cortex (e.g., parietal cortex) [43, 44], despite the preliminary hypothesis of a selective ventromedial prefrontal cortex dysfunction in PG patients [42], our data would suggest a more generalized frontal lobe impairment. This is also supported by a recent study showing alteration of both dorsolateral prefrontal and orbitofrontal cortex in PG [45]. However, it is still unclear whether the observed frontal lobe abnormalities should be considered a primary phenomenon linked to the aetiology of PG, or secondary to some symptomatological features, or to the comorbid psychopathological conditions.

Taken together, our results would seem to confirm an altered functioning of the prefrontal areas which would determine a cognitive rigidity that might be expressed through altered executive functions and the decision making process; this could represent a factor of vulnerability to the development of impulsive and/or compulsive behaviours, such as those typical of PG. It would be interesting to replicate these findings in patients suffering from PG alone, or with distinct comorbidity patterns.