Multiple asynchronous stimulus‐ and task‐dependent hierarchies (STDH) within the visual brain's parallel processing systems

Abstract

Results from a variety of sources lead ineluctably to a re-appraisal of the hierarchical and parallel processing strategies used by the brain to construct an image of the visual world. Contrary to common supposition, there are at least three "feed-forward" anatomical hierarchies that reach the primary visual cortex (V1) and the visual areas outside it, in parallel. These anatomical hierarchies do not conform to the temporal order with which visual signals reach and activate V1 and the specialized visual areas.Furthermore, neither the anatomical hierarchies nor the temporal order predict the perceptual hierarchies. The latter shows that we see (and become aware of) different visual attributes at different times, with colour leading form (orientation) and visual motion, even though it is signals from fast moving stimuli that are the earliest to reach the visual cortex (of area V5). Parallel processing, on the other hand, is much more ubiquitous than commonly supposed, but is subject to a barely noticed but fundamental aspect of brain operations, namely that different parallel systems operate asynchronously with respect to each other and reach perceptual end-points at different times.This re-assessment leads to the conclusion that the visual brain is constituted of multiple, asynchronously operating, task-and stimulus-dependent parallel hierarchies even within systems sub-serving a single attribute such as form or visual motion; which of these parallel anatomical hierarchies has temporal and perceptual precedence at any given moment is stimulus-and task-related and critically dependent upon the visual brain's ability to undertake multiple operations asynchronously.

Pre-stimulus feedback connectivity biases the content of visual experiences

“…The early latency of our gamma effect is in line with recent MEG work showing differential 40ms evoked responses to forms (59). The 40 ms response was also reported for face stimuli and localized in FFA and other visual areas (60), indicating parallelism in visual input streams and challenging the commonly held supposition that V1 is the sole entry point for visual input (61,62). That our gamma cluster starts at stimulus onset, and is to some extent observable even before, suggests that it likely reflects an interaction between pre-stimulus anticipatory processes (in which gamma oscillations have also been implicated; eg: (14,63,64)) and very early sensory processes.…”

The Bayesian‐Laplacian brain

“…Hence a strictly hierarchical organization for form based on a single feed‐forward system from the lateral geniculate nucleus through V1 (as is almost universally supposed), and in which cells within the brain's form system acquire increasingly more complex properties, enabling them to respond to complex objects and faces, is probably unlikely. Rather, there appears to be multiple hierarchical systems, which operate in parallel—with recurrent message passing for belief updating and which are time and context dependent (Zeki, ). Hence, if the OS cells constitute the biological prior from which forms and objects are constructed in the brain, one must entertain the possibility that there may be several different kinds of OS cells, embedded in different visual areas, each of which may constitute a biological or β prior for generating different kinds of posteriors, some of them artefactual.…”

The Bayesian-Laplacian Brain

“…Hence a strictly hierarchical organization for form based on a single feed-forward system from the lateral geniculate nucleus through V1 (as is almost universally supposed), and in which cells within the brain's form system acquire increasingly more complex properties, enabling them to respond to complex objects and faces, is probably unlikely. Rather, there appears to be multiple hierarchical systems, which operate in parallel and which are task and stimulus dependent (Zeki, 2016). Hence, if the OS cells constitute the biological prior from which forms and objects are constructed in the brain, one must entertain the possibility that there may be several different kinds of OS cells, embedded in different visual areas, each of which may constitute a biological or β prior for generating different kinds of posteriors, some of them artifactual.…”