The primate dopamine system is involved in appetitively motivated behaviours, including certain forms of learning, for example, visual discrimination learning. Furthermore, food restriction in animals and anorexia in humans is associated with impaired dopamine signaling. Based on this, we hypothesized that patients with anorexia nervosa (AN) would show a deficit in visual discrimination learning. In a dynamic categorization task involving the learning of a series of two-alternative forced-choice visual discriminations, conceptually identical to one shown to activate dopamine neurons in primates, and sensitive to dopaminergic manipulations in humans, patients with AN showed a deficit in learning that was most pronounced in the early stages of acquisition. In contrast, AN showed spared performance on a pattern recognition memory test sensitive to medial temporal lobe lesions, but insensitive to dopaminergic manipulations. We conclude that impaired appetitive function in patients with AN extends to include deficits in visual discrimination learning, and that this deficit represents indirect evidence for altered dopaminergic neurotransmission in AN.
The dopamine system has long been implicated in appetitive behaviour, including the processing of food and stimuli that reliably predict the presentation of food (Wise and Rompre, 1989, Robbins and Everitt, 1996, Hoebel et al., 1999, Ikemoto and Panksepp, 1999 and Berridge, 2001). Recent experiments have helped to refine this role, indicating a specific role for dopamine in coding reward prediction errors, critical for certain forms of learning (Berridge, 2001 and Schultz et al., 1997). For example, Hollerman and Schultz (1998) have shown that during visual discrimination learning, dopamine neurons reflect the changes in reward (or more generally, outcome) prediction during individual learning episodes; they are activated by rewards during early trials, when errors are frequent and rewards unpredictable, but activation progressively reduces as performance is consolidated and rewards become predictable (i.e. as a learning set develops). These, and other findings have lead to the hypothesis that dopamine mediates activity in a ‘behavioural approach’ (Pickering and Gray, 2001 and Gray 1987); ‘behavioural facilitation’ (Depue and Collins, 1999) or ‘Seeking’ (Panksepp, 1982 and Panksepp, 1998) system involved in exploratory/investigative behaviors critical for the acquisition of valued biological resources.
Dopamine lesions lead to weight loss and anorexia in animals (Zigmond and Stricker, 1972, Baez et al., 1977 and Szczypka et al., 2001) and food restriction leads to altered dopamine system function (Pothos et al., 1995 and Kosta et al., 1999). In humans, patients with anorexia nervosa (AN) have impaired dopamine function, as indexed for example by reduced CSF concentrations of the dopamine metabolite homovanillic acid, and altered growth hormone response to apomorphine stimulation (Barry and Klawans, 1976, Gillberg, 1983, Owen et al., 1983, Kaye et al., 1999 and Brambilla et al., 2001). We hypothesized that, if dopamine is critical for error-driven learning, and anorexia is associated with impaired dopamine signaling, then individuals with AN should show an impairment in such learning.
In the current study, we examined the performance of patients with AN and healthy controls on a task comprising a suite of two-alternative forced-choice visual discriminations and reversals (Roberts, Robbins and Everitt, 1988). This task is sensitive to manipulations of dopaminergic neurotransmission, including l-dopa treatment in Parkinson's disease (Lange et al., 1992), and is analogous to discrimination learning tasks shown to activate dopamine neurons in primates (Hollerman and Schultz, 1998). Given our hypothesis, we predicted that patients with AN would be impaired in such learning, perhaps especially during the early stages of the task, when dopaminergic activity should be at a maximum (Hollerman and Schultz, 1998). As a control task, we included a test of pattern recognition memory requiring the detection of stimulus repetitions. This task is sensitive to damage to medial temporal lobe structures (Owen et al., 1995), but is insensitive to dopaminergic manipulations (Lange et al., 1992 and Mehta et al., 1999). We predicted no deficit on this task in AN.
