# Specific mechanisms linking network information processing to the generation of qualia

## TL;DR

This paper explores how neural networks process information and how this might lead to the experience of consciousness and subjective feelings.

## Contribution

The paper introduces a novel perspective on how neural networks generate self-referential representations that could underlie qualia.

## Key findings

- Neural networks can identify and represent spatially distributed input patterns.
- Direct feedback of these representations may lead to self-referential identity acquisition.
- Top-down modulation may determine which qualia are experienced at any given moment.

## Abstract

There have been many very promising theories published concerning the generation of consciousness. These theories mostly link the emergence of consciousness to neural activity, but very few attempt to show how that neural activity specifically causes experience to occur. This article explores this problem at the level of individual networks by examining the information changes that occur as input patterns are processed. It looks at how networks can identify spatially distributed input patterns, and generate representations of that identity. It argues that if those representations are directly fed back then such networks will be identifying their own depictions of the original identity. It goes on to argue that in this state the identity acquired is not what the input is to the network but how the input seems to it. There would be content to that inner portrayal that must present itself in some way to the receiver, and this could underlie the emergence of qualia. The article goes on to argue how top-down modulation could select which qualia are established at any moment.