All models are wrong, but some are useful (Box, Reference Box, Launer and Wilkinson1979)

Psychosis remains one of the most distressing, yet intriguing, human conditions. The profession of psychiatry grew up in asylums trying to grapple with psychosis over a century ago. How best to understand and treat psychosis is still an ongoing academic challenge. At one level psychosis is a behavioural phenomenon. At another it is tied to the brain, its chemistry and circuits. Thus, any complete explanation of psychosis will have to try and link explanations at these two levels of analysis: a neurobiological/neurochemical level and the psychological/behavioural level. A few years ago one of us proposed that the concept of ‘motivational salience’ may be one such linking variable (Kapur, Reference Kapur2003). The link between dopamine and the limbic system and the concept of salience is strong. But, this link is built mainly on animal studies using a variety of rather simple conditioning paradigms using rewarding and aversive stimuli. To make these concepts applicable to schizophrenia, an anthropomorphized notion of salience was proposed. Salience was defined as a process whereby objects and representations, through the process of association, come to be attention-grabbing and capture thought and behaviour. While this has served as a useful general heuristic – how does one test such a broad notion? Over the past few years several innovative groups has developed interesting and novel ways to do this and Roiser et al. provide one such interesting approach in this issue (Roiser et al. Reference Roiser, Stephan, den Ouden, Barnes, Friston and Joyce2008).

Roiser and colleagues used a probabilistic monetary task where adaptive and aberrant salience were operationalized as outcome measures, both having explicit and implicit components. They used medicated patients with schizophrenia and healthy matched controls and showed that the patients exhibited reduced adaptive salience. These were the hypothesized results because these were medicated patients and it was hypothesized that their treatment would have corrected their aberrant salience, but, dampened their adaptive salience. Interestingly however, Roiser et al. (Reference Roiser, Stephan, den Ouden, Barnes, Friston and Joyce2008) also found that within these treated patients, those with more positive symptoms showed increased aberrant salience compared to patients with less symptom load but also that this finding was mainly driven by the presence of delusions. Although these findings warrant further studies in unmedicated patients, using a probabilistic monetary task and making some sophisticated operationalizations the authors showed results that are predicted by the aberrant motivational salience model on the psychological level.

While Roiser et al. focused on a behavioural manifestation of salience, several recent studies have focused on the neurobiological level. The ventral striatum plays a central role in reward processing and it has been suggested that activations of the ventral striatum normally mediate motivational salience of environmental stimuli. A functional magnetic resonance imaging (fMRI) study by Juckel and collaborators (Juckel et al. Reference Juckel, Schlagenhauf, Koslowski, Wustenberg, Villringer, Knutson, Wrase and Heinz2006b) using a monetary incentive task showed that, compared to controls, unmedicated patients with schizophrenia displayed a reduced activation in the ventral striatum for cues predicting reward relative to a cue predicting no gain. This reduced activation was normalized when patients were treated with atypical medication but not with typical antipsychotics as evidenced by both a cross-sectional (Juckel et al. Reference Juckel, Schlagenhauf, Koslowski, Filonov, Wustenberg, Villringer, Knutson, Kienast, Gallinat, Wrase and Heinz2006a) and a longitudinal cross-over study (Schlagenhauf et al. Reference Schlagenhauf, Juckel, Koslowski, Kahnt, Knutson, Dembler, Kienast, Gallinat, Wrase and Heinz2008). The results suggest that the properties of the newer antipsychotics, such as lower striatal D₂ blockade, modulation by serotonin and other systems, or their faster dissociation from the receptor, are beneficial for motivational salience which is in line with the normalized aberrant salience results of medicated patients. Interestingly, in the study of Roiser et al. 88% of the patients took second-generation drugs (Roiser et al. Reference Roiser, Stephan, den Ouden, Barnes, Friston and Joyce2008).

Using a different approach with fMRI, Murray et al. (Reference Murray, Corlett, Clark, Pessiglione, Blackwell, Honey, Jones, Bullmore, Robbins and Fletcher2008) undertook a study of first-episode psychosis patients, and found a disruption in reward prediction error among the patients. Abnormal activations in dopaminergic regions, such as ventral tegmental area/substantia nigra and the striatum were observed and the patients also had difficulties in discriminating between motivationally salient and neutral stimuli. Using an aversive conditioning paradigm with a loud noise as unconditioned stimulus in medicated patients, we found a similar abnormality in discriminating between motivational salient and neutral stimuli using both self-report data and autonomic measures (Jensen et al. Reference Jensen, Willeit, Zipursky, Savina, Smith, Menon, Crawley and Kapur2008). This was accompanied with deviant activation patterns in the ventral striatum suggesting that patients activated this region more towards neutral stimuli than healthy controls while they showed more similar activations to stimuli that the controls found motivationally salient, i.e. the conditioned stimulus. Both these studies (Jensen et al. Reference Jensen, Willeit, Zipursky, Savina, Smith, Menon, Crawley and Kapur2008; Murray et al. Reference Murray, Corlett, Clark, Pessiglione, Blackwell, Honey, Jones, Bullmore, Robbins and Fletcher2008) thus showed evidence of abnormal assignment of motivational salience and abnormal activation patterns in the hypothesized regions in patients with psychosis.

The behavioural and imaging data reported above are consistent with the idea that patients aberrantly assign motivational salience to neutral stimuli (Kapur, Reference Kapur2003), and this process may be one of the aberrations that predisposes them to psychosis. However, the different studies have used a slightly different interpretation and operationalization of ‘salience’. This is not surprising by itself, because concepts like ‘reward’ itself are operationalized in several different ways, and the concept of salience is loosely defined making such interpretations justified. The paper by Roiser et al. (Reference Roiser, Stephan, den Ouden, Barnes, Friston and Joyce2008), and several other recent ones are encouraging, but, the real challenge, as the title of this commentary suggests, is whether these ways of looking at psychosis will help us understand psychosis at a deeper level or deal with it more effectively. For that, more studies are needed: to replicate these initial results, to extend these studies to drug-naive patients, to examine the effect of medications in longitudinal studies and to look at the predictive value of these findings.