About this Research Topic
Embodied perception-action dynamics are influenced by affordances. Affordances are environmental properties, such as height, shape, weight, etc., that can allow opportunities for various actions in relation to the mobility of the perceiver. There are additional environmental factors that can influence perception. Temporal properties, for example, speed, rhythmicity, and interstimulus intervals are known to influence perception. Another factor that may affect perception is the internal state of the organism, for example, attention. Perception and action play important roles in the interaction of the brain with the environment. Since the brain has no previous knowledge of the events before they unfold, there are processes that involve the prediction of information and computation of surprise. Moreover, the interaction with the environment increases the mutual information in the brain, given the occurrence of sensory events in the surroundings. It was proposed that the increase in surprisal information and mutual information may provide a basis for perception and action.
Furthermore, unsupervised models of information processing in the brain, such as active inference have been gaining ground in recent decades. More recently, variational autoencoders (VAEs), a type of stochastic unsupervised learning has been trained to detect low-dimensional features of images in the latent variable space, for instance face pose or smile. Data, such as fMRI, MEG signals and spikes patterns during specific cognitive tasks can be trained by the VAEs to estimate latent variables. Latent variables may represent the low-dimensional reconstruction of the features of the external stimuli, like possible actions with respect to external objects (affordance). Therefore, these latent variables may be analyzed for their association with various actions of a task to shed light on information processing during cognitive tasks.
This Research Topic seeks contributions from researchers working in different disciplines, which will shed light on the role of the environmental factors, like affordance, temporal properties and surprisal information in perception and action. Papers may use various computational approaches, such as unsupervised learning to understand the information processing in the brain. Manuscripts based on clinical cases, imaging studies, molecular studies and psychophysics are also welcome.
Keywords: Variational auto-encoders, binding, unsupervised learning, Shannon information, encoding/decoding, predictive coding
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.