Precision or gain control within the brain’s body model

When controlling action—e.g. hand movements—the brain typically relies on seen and felt posture information to represent ‘its’ body. It is thought that the brain combines these estimates into a multisensory (e.g. hand) representation in a probabilistic fashion, accounting for how reliable each estimate is in the given context. I will present behavioural, brain imaging, and computational modelling work—which jointly suggest that during action, the weights (i.e., the precision) assigned to visual vs proprioceptive information about body position can be changed in a ‘top-down’ fashion by e.g. adoption of an ‘attentional task set’. Such a contextualising function of attention on sensory evidence (a.k.a. gain control) is in line with predictive coding models of body representation, along which the expected precision of sensory prediction errors is augmented or attenuated—depending on the current context—based on internal model beliefs. I will conclude that to some degree body representations can be deliberately ‘shaped’, and will discuss what this implies for the near future—in which entertaining and switching between multiple body representations, e.g. in virtual realities, may be increasingly common.