Psychiatric disturbance in Parkinson’s Disease (PD), particularly visual hallucinations (VHs), can be a frequent and disturbing complication of the disease.1 Hallucinations occur in patients with PD with a prevalence ranging from 8% to 40%2 and are associated with behavioural and functional problems and higher mortality.3 Visual hallucinations have been reported with all anti-Parkinson’s medications4 and it is estimated that up to 33% of patients with PD undergoing long-term treatment will have VHs during the course of their illness.5 Nevertheless, comparisons of hallucinators and non-hallucinators have seldom shown major differences in drug history5 and VHs have been reported in patients with PD before any drugs have been taken.6

Several lines of evidence have emerged indicating that hallucinations are closely linked to cognitive impairment, reflected by a general degradation in information-processing abilities.7 A recent clinical study suggests that patients with PD with VHs show reduced performance on tasks that explore executive functioning when compared with non-hallucinators, suggesting that executive dysfunction may be considered to be a risk factor for the development of hallucinations in non-demented patients with PD.8 We report our findings on tests of executive function, such as tests of inhibitory ability, short-term memory and working memory on patients with PD with and without hallucinations and age-matched controls.

METHODS

Participants

All patients had a presumptive clinical diagnosis of PD with (n = 17) and without VHs (n =  20). Selection of participants was carried out with the help of neurologists and ophthalmologists to ensure that no participant had a clinical diagnosis of Alzheimer’s disease or Lewy body dementia, or had any current or previous history of eye disease. Normal controls (n = 20) were matched for age.

Tests of inhibitory ability

A standard colour Stroop task was administered.9 The Stroop task consisted of three 45 s trials during which participants read out as many words as possible. Two control trials (trial 1 and trial 2) were performed to allow for the individual differences in reading speed and colour naming. In trial one (T1), the words red, blue and green were written in black ink. In Trial 2 (T2), participants were asked to name the colour of ink present on cards on which a series of Xs were written. In trial 3 (T3), participants were asked to name the colour of ink in which a colour word was written. The Stroop interference score was computed using the formula 100 + I − (T1 × T2)/(T1 + T2). This formula subtracted a “predicted interference” score (T1 × T2)/(T1 + T2) from the raw score I,9 with 100 added to avoid negative numbers.

Go/no-go task10

Participants viewed a computer screen with two buttons in front of them: a “near” button and a “far” button. Instructions then appeared on the screen as commands “Stay” or “Move”. On “Move”, participants had to lift their finger from the near button, depress the far button and then return their finger to the near button. On “Stay”, participants did not lift their finger from the near button.

Category fluency11

Participants were asked to generate as many words as possible in 60 s from six categories (furniture, fruit, animals, tools, items of clothing and insects) and were told not to give the same item more than once.

Tests of working memory

2-back task12

Four-letter abstract nouns were presented on-screen one at a time at a rate of one word every 2 s. Participants were asked to press one button whenever the current stimulus matched the stimulus that had appeared two before it and to press a second button whenever the stimulus was not a match to the stimulus that occurred two before.

Reading span13

A sentence was presented on a computer screen, which the participants read aloud. They were then asked to answer a simple comprehension question about the sentence. This was then repeated with another sentence. The participants were then asked to recall the final word of each of the sentences. Participants performed five sets of these two-sentence presentations.

Tests of short-term memory

Digit span (WAIS-III)

Digit strings were presented orally at a rate of one digit per second. Participants were asked to repeat the digits in the same order as they had been presented. Two attempts were allowed at each digit string length.

Word span14

Task administration and scoring were identical to that of Digit Span. The longest string administered was nine words. All words were one-syllable, concrete nouns.

RESULTS

Participant demographics and comparison of tests are summarised in table 1. The results were analysed using analysis of variance (ANOVA) with post-hoc pair-wise Tukey’s test to explore significant main effects.

Tests of inhibition

Analyses of variance (ANOVA) indicated significant group effects on all tests of inhibitory ability (see table 1). Stroop inference scores produced significant group effects [F (2, 56) = 10.10, p<0.01], with the hallucinating patients performing the worst and normal controls the best. On the go/no-go task, the ANOVA indicated a group effect [F (2, 56) = 17.97, p<0.001], with the same pattern of results as the Stroop task. On fluency perseverations, the group effect [F (2, 56) = 40.82, p<0.01] indicated that hallucinators made more perseverations than non-hallucinators and normal controls ([t (35) = 3.28, p<0.01] and [t (38) = 8.36, p<0.01], respectively).

Tests of working memory

The ANOVA conducted on the corrected hit rate scores indicated a significant effect of group [F (2, 56) = 29.9, p<0.01], with both PD groups performing less well than the normal control group. There was an effect of group for the percentage of hits [F (2, 56) = 27.40, p<0.01], with the PD groups performing less well than the control group. A different pattern was evident with false alarm data. There was an effect of group [F (2, 56) = 7.63, p<0. 01], but here the hallucinating group performed worse than the non-hallucinating group [t (35) = 2.33, p<0.05], who in turn performed less well than normal controls [t (38) = 3.79, p<0.01]. For patients with PD with VHs, the false alarm scores correlated with all measures of inhibition (stroop score, r = −0.54, p<0.05; go/no-go, r = −0.6, p<0.05; fluency perseverations, r = 0.48, p<0.05). In contrast, percentage hit rate correlated with digit span (r = −0.51, p<0.05). Unlike the group with VHs, the false alarm scores of the other two groups did not show significant correlations with measures of inhibitory ability (r<.25, p >.25). On the reading span task, the analyses indicated a significant effect of group [F (2, 56) = 20.15, p<0. 01], with the PD groups performing less well than controls.

Tests of short-term memory

On digit span and word span, the PD groups performed less well than controls ([F (2, 56) = 6.83, p<0. 01] and [F (2, 56) = 6.59, p<0. 01], respectively).

DISCUSSION

Patients with VHs were impaired on all three tests of inhibitory ability when compared with the patients with PD without VHs and controls. In addition, the same pattern of results was seen in the false alarm rates in the 2-back task; the patients with VHs performed worse than patients without VHs and controls. The difference highlighted on inhibitory ability might simply be a matter of disease severity or duration of illness; however, if this were the case, working memory would be expected to be equally disrupted. Furthermore, the more severe motor impairment present in the hallucinating patient group and their subsequent speed of response may be a greater contributor to their lower scores in control trials of the Stroop task, especially when considered together with the comparable patient scores on the reading span task.

Inhibition-like deficits have been linked to a disruption of striatal-dopaminergic projections that are thought to be particularly important for selective attention.15 16 Indeed, the recent PAD (Perception and Attention Deficit) model of VHs proposes that a combination of deficits in attentional binding and object perception is essential to the occurrence of recurrent complex VHs.7 In addition, the HEAR (Hallucinatory Experience of Auditory Representation) model incorporates a contextual binding deficit, as well as an inhibitory control deficit.17

However, although inhibitory ability might indeed represent an important role in the phenomenology of VHs, this does not imply that the proposed inhibitory disturbances are alone in producing the hallucinatory experience. Our previous work found that patients with PD with VHs had intact visual imagery, but poor object perception and “switching” between modalities at study and test stages,18 a task which would require an inhibitory process. Together with the current observation that inhibitory mechanisms may be implicated, these findings serve to highlight the complexity of the processes involved and perhaps the need to characterise the hallucinating individual more precisely.

Deficits in executive functions in patients with PD are well documented,19 although few studies have directly investigated working memory deficits and inhibitory processes in hallucinating patients. Here, the dissociation in inhibitory processes and working memory could suggest differential depletion of dopamine in patients with PD with and without hallucinations, and anatomically points to frontal dysfunction in the generation of VHs and further contributes to the integrative model of VHs in PD.