Fueled by developments in computational neuroscience, there has been increasing interest in the underlying neuro-computational mechanisms of psychosis. One successful approach involves predictive coding and Bayesian inference. Here, inferences regarding the current state of the world are made by combining prior beliefs with incoming sensory signals. Mismatches between prior beliefs and incoming signals constitute prediction errors that drive new learning. Psychosis has been suggested to result from a decreased precision in the encoding of prior beliefs relative to the sensory data, thereby garnering maladaptive inferences. Here, we review the current evidence for aberrant predictive coding and discuss challenges for this canonical predictive coding account of psychosis. For example, hallucinations and delusions may relate to distinct alterations in predictive coding, despite their common co-occurrence. More broadly, some studies implicate weakened prior beliefs in psychosis, and others find stronger priors. These challenges might be answered with a more nuanced view of predictive coding. Different priors may be specified for different sensory modalities and their integration, and deficits in each modality need not be uniform. Furthermore, hierarchical organization may be critical. Altered processes at lower levels of a hierarchy need not be linearly related to processes at higher levels (and vice versa). Finally, canonical theories do not highlight active inference – the process through which the effects of our actions on our sensations are anticipated and minimized. It is possible that conflicting findings might be reconciled by considering these complexities, portending a framework for psychosis more equipped to deal with its many manifestations.

Patients suffering from schizophrenia may report unusual experiences of their own actions. They may either feel that external forces are controlling their actions or even their thoughts, or they may feel in control of events that in fact are not caused by their actions. Most theories link these disturbances in the sense of agency to deficits in motor prediction, resulting in a mismatch between predicted and actual sensory feedback at a central comparator mechanism. Such theories therefore can account for situations in which the sense of agency is reduced. However, other experiments as well as clinical observations show an enhanced rather than reduced sense of agency in schizophrenic patients. Here, we distinguish between a predictive and a retrospective mechanism where both contribute to the experience of agency, and show that schizophrenia is associated with a specific impairment to the predictive component. We measured subjective time estimates of self-initiated voluntary action (a key press) that were followed by a sensory effect (a tone). When the voluntary actions had a high probability of causing tones, healthy volunteers showed a predictive shift of the perceptual estimate of the action towards the tone, even on occasional trials where the tone was omitted. No such shift occurred in the absence of the tone on blocks when tones were less frequent. The predictive component of action awareness was calculated as the difference between time estimates on 'action only' trials from blocks with lower and higher tone probabilities. Schizophrenic patients lacked this predictive component of action awareness, showing a shift on 'action only' trials, regardless of the probability of the tone. Importantly, the schizophrenic deficit in predicting the relation between action and effect was strongly correlated with severity of positive psychotic symptoms, specifically delusions and hallucinations. Furthermore, the patients showed an exaggerated retrospective binding between action and tone, shifting the perceived time of action whenever the tone occurred, relative to when it did not occur. Our quantitative, implicit measures show how basic sensory and motor experience may be altered in acute psychosis. The enhanced sense of agency in schizophrenia reflects reliance on retrospection, rather than prediction, to associate actions with external events. The failure to predict the effects of one's own actions may underlie the blurring and confusion in the relationship between the self and the world that characterizes acute psychosis.

The experience of agency, i.e., the registration that I am the initiator of my actions, is a basic and constant underpinning of our interaction with the world. Whereas several accounts have underlined predictive processes as the central mechanism (e.g., the comparator model by C. Frith), others emphasized postdictive inferences (e.g., post-hoc inference account by D. Wegner). Based on increasing evidence that both predictive and postdictive processes contribute to the experience of agency, we here present a unifying but at the same time parsimonious approach that reconciles these accounts: predictive and postdictive processes are both integrated by the brain according to the principles of optimal cue integration. According to this framework, predictive and postdictive processes each serve as authorship cues that are continuously integrated and weighted depending on their availability and reliability in a given situation. Both sensorimotor and cognitive signals can serve as predictive cues (e.g., internal predictions based on an efferency copy of the motor command or cognitive anticipations based on priming). Similarly, other sensorimotor and cognitive cues can each serve as post-hoc cues (e.g., visual feedback of the action or the affective valence of the action outcome). Integration and weighting of these cues might not only differ between contexts and individuals, but also between different subject and disease groups. For example, schizophrenia patients with delusions of influence seem to rely less on (probably imprecise) predictive motor signals of the action and more on post-hoc action cues like e.g., visual feedback and, possibly, the affective valence of the action outcome. Thus, the framework of optimal cue integration offers a promising approach that directly stimulates a wide range of experimentally testable hypotheses on agency processing in different subject groups.

Our nervous system continuously combines new information from our senses with information it has acquired throughout life. Numerous studies have found that human subjects manage this by integrating their observations with their previous experience (priors) in a way that is close to the statistical optimum. However, little is known about the way the nervous system acquires or learns priors. Here we present results from experiments where the underlying distribution of target locations in an estimation task was switched, manipulating the prior subjects should use. Our experimental design allowed us to measure a subject's evolving prior while they learned. We confirm that through extensive practice subjects learn the correct prior for the task. We found that subjects can rapidly learn the mean of a new prior while the variance is learned more slowly and with a variable learning rate. In addition, we found that a Bayesian inference model could predict the time course of the observed learning while offering an intuitive explanation for the findings. The evidence suggests the nervous system continuously updates its priors to enable efficient behavior.

Voluntary actions typically produce suppression of afferent sensation from the moving body part. We used transcranial magnetic stimulation to delay the output of motor commands from the motor cortex during voluntary movement. We show attenuation of sensation during this delay, in the absence of movement. We conclude that sensory suppression mainly relies on central signals related to the preparation for movement and that these signals are upstream of primary motor cortex.Self-generated actions often lead to an attenuation of sensation from the moving body part1 and from other parts of the body that are actively contacted2,3. The underlying neural mechanism may involve efference copy attenuating movement-related sensations4,5. Recent results have suggested an important peripheral component to this attenuation: during active wrist movements in the primate, cutaneous inputs are presynaptically inhibited at the spinal cord afferents before movement6. Other studies have demonstrated attenuation centrally within the brain. For example, the amplitude of cortical somatosensory-evoked potentials (SEPs) or somatosensory-evoked magnetic fields (SEFs) is reduced before the onset of movement, before any peripheral feedback7-9. However, SEP attenuation is often dissociated from perception: changes in sensory perception may occur without the modulation of SEPs, and vice versa10. Therefore, it remains unclear how and to what extent central signals influence subjective perception during movement. We therefore investigated whether earlier stages of the motor hierarchy, which prepare motor commands before their dispatch to the spinal cord, contribute to the attenuation of sensory inputs. An ideal, if somewhat artificial, situation to address this question would involve a motor command being generated centrally, but its dispatch from the brain to the spinal cord being blocked or delayed. Transcranial magnetic stimulation (TMS) over primary motor cortex during voluntary movement induces such a delay in corticospinal output without otherwise affecting the motor pattern11. This delay seems to be largely central in origin: spinal excitability increases12, or remains unchanged13, rather than decreases during the TMSinduced delay period. We used this paradigm to investigate whether attenuation would occur after the preparation of motor commands and before their dispatch to the periphery12.Correspondence should be addressed to M.V. (martin.voss@charite.de). Note: Supplementary information is available on the Nature Neuroscience website. COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests. To measure sensory suppression, we applied brief electrical cutaneous stimuli simultaneously to left and right index fingers of 21 subjects (informed written consent was obtained; for details, see Supplementary Methods online). The left finger served as a reference, remaining at rest and receiving a fixed stimulus intensity throughout. The stimulus to the right finger was varied from trial to trial. We...

When a part of the body moves, the sensation evoked by a probe stimulus to that body part is attenuated. Two mechanisms have been proposed to explain this robust and general effect. First, feedforward motor signals may modulate activity evoked by incoming sensory signals. Second, reafferent sensation from body movements may mask the stimulus. Here we delivered probe stimuli to the right index finger just before a cue which instructed subjects to make left or right index finger movements. When left and right cues were equiprobable, we found attenuation for stimuli to the right index finger just before this finger was cued (and subsequently moved). However, there was no attenuation in the right finger just before the left finger was cued. This result suggests that the movement made in response to the cue caused ‘postdictive’ attenuation of a sensation occurring prior to the cue. In a second experiment, the right cue was more frequent than the left. We now found attenuation in the right index finger even when the left finger was cued and moved. This attenuation linked to a movement that was likely but did not in fact occur, suggests a new expectation-based mechanism, distinct from both feedforward motor signals and postdiction. Our results suggest a new mechanism in motor-sensory interactions in which the motor system tunes the sensory inputs based on expectations about future possible actions that may not, in fact, be implemented.

Humans and other primates demonstrate an exquisite ability to precisely shape their hand when reaching out to grasp an object. Here we used a recently developed transcranial magnetic stimulation paradigm to examine how information about an object's geometric properties is transformed into specific motor programs. Pairs of transcranial magnetic stimulation pulses were delivered at precise intervals to detect changes in the excitability of corticocortical inputs to motor cortex when subjects prepared to grasp different objects. We show that at least 600 ms before movement, there is an enhancement in the excitability of these inputs to the corticospinal neurons projecting from motor cortex to the specific muscles that will be used for the grasp. These changes were objectand muscle-specific, and the degree of modulation in the inputs was correlated with the pattern of muscular activity used later by individual subjects to grasp the objects. In a number of control experiments, we demonstrated that no change in excitability was observed during object presentation alone, under conditions in which subjects imagined grasping the object, or before movements involving the same muscles but without an object. This finding demonstrates a cortico-cortical mechanism subserving the transformation from the geometrical properties of an object to the outputs from motor cortex before grasp that is specific for objectdriven movements.cortex ͉ transcranial magnetic stimulation ͉ I-wave ͉ corticospinal

No abstract