What we perceive depends on how much it costs us.

Interesting work from Hagura et al. showing that our perceptual decisions are biased by the action costs that are associated with our subsequent decisions is summarized by de Lange and Fritsche:

Perceptual decision-making is not solely determined by the characteristics of the sensory stimulus, but is influenced by several factors such as expectation, reward, and previous history , which may all facilitate perceptual decision-making under uncertainty. A factor that has been mostly neglected in laboratory settings is that, in everyday life, making perceptual decisions between several options entails actions which can differ dramatically in their associated motor cost. For instance, imagine standing in front of an apple tree and searching for the reddest-looking apple to pick. Naturally, picking apples higher up in the tree requires more physical effort than picking low-hanging apples. Therefore, your decision about whether to pick a high- or low-hanging apple has consequences for the subsequently accruing motor costs. Does this difference in expected motor costs influence your perceptual decision about the color of the apples? That is, do you judge the low-hanging fruit as more red? We know that motor costs can bias the choice behavior in perceptual decision-making tasks to maximize the expected utility of the choice, but it is unclear whether motor costs can affect the perceptual decision itself.

Hagura et al. shed light on this question. They asked participants to indicate the direction of motion (leftward or rightward) of a cloud of moving dots, by moving one of two robotic manipulanda with their left or right hand, respectively. Unknown to the participants, the resistance for moving one of the manipulanda was gradually increased, while the other remained unchanged. In line with a previous study, Hagura et al. found that participants subsequently showed a tendency to avoid decisions for the motion direction that was associated with the energetically more-costly motor response. Crucially, after the induction of asymmetric motor response costs for manual responses, participants showed a similar bias when indicating their decisions vocally. This transfer of the bias onto decisions reported with a different effector – for which motor response costs were not manipulated – suggests that the repeated exposure to motor response costs associated with a particular decision can bias future perceptual decisions themselves. Thus, the manual-to-vocal transfer effect provides first evidence that motor costs are not necessarily integrated with perceptual decisions at the motor output stage, but that recent experience of motor costs can change how sensory input is transformed into a perceptual decision.

The current results suggest that motor costs can bias perceptual decisions before they are transformed into an effector-specific response. However, the exact stage along the visual processing stream at which this bias occurs is unclear. Motor costs could target an early stage of visual processing, biasing the sensory representation of visual input, or occur at a later stage, targeting a general, effector-unspecific decision stage. Using a drift-diffusion model approach, the authors found that their data were best explained by a model in which the motor costs change the decision bound that is used to make the decision, rather than the evidence accumulation process itself. This suggests that motor costs target a later decision layer, rather than the sensory representation, and distinguishes it from other processes such as attentional biases which affect the accumulation rate of sensory evidence. An intriguing outstanding question, related to this issue, is whether motor costs can alter the appearance of visual stimuli