Anchoring visual subjective experience in a neural model: The coarse vividness hypothesis

Abstract

Subjective experience often accompanies perception and cognition. This elusive feeling is difficult to characterize, both theoretically and experimentally. Perceptual subjective experience is at the heart of a theoretical debate in consciousness research: does it correspond to a genuine psychological and biological process independent from cognitive abilities, or is it a cognitive illusion, a post-hoc construct, implying that perceptual consciousness can be reduced to a sum of cognitive functions? We reconsider this debate in the light of known properties of the visual system, derived from studies on visual object and scene recognition but not specifically targeting consciousness issues. We propose here that initial visual subjective experience is characterized by two key properties, coarseness and vividness: initial subjective experience is integrated, meaningful, but does not contain detailed information. Subjective experience is likely to arise first in high-level visual areas, in which information is encoded in a coarse and integrated manner. We propose that initial subjective experience is related to the concept of “vision at a glance”, thought to result from a fast, implicit feed-forward sweep of activity in the visual system progressing from low-level areas to high-level areas (Hochstein and Ahissar (2002) Neuron, 36, 791–804). The details needed to overtly guide behavior would be retrieved in a secondary processing step of “vision with scrutiny”, proceeding in a feed-back manner, from high-level to low-level areas. This secondary and optional descending process could thus later enrich conscious visual percepts with details. Our hypothesis provides parsimonious explanations for two intriguing findings: the double dissociation between attention and consciousness, and the mismatch between objective measures and subjective reports, that is sometimes used to argue that subjective experience is an illusion. We argue here that because visual subjective experience is initially coarse, it should not be probed by asking subjects to specify details. The coarse vividness hypothesis therefore offers a framework that accounts for the existence of an initial genuine subjective experience, defined by its coarseness and vividness, optionally followed by more refined and detailed processing that could underlie finer perceptual and cognitive abilities.

Introduction

Perceptual subjective experience refers to the way the world appears to us via our senses. It is an intuitive notion, fundamentally constitutive of our human nature. We all share the intuition that a robot, however smart, is lacking any feeling associated with the complex operations that it can execute: a robot is not human, therefore it lacks subjective experience. Subjective experience nevertheless remains an elusive notion, difficult to frame in a scientific theory, since it is essentially a private experience, that is not easily accessible to the experimenter. Nagel (1974) famously illustrated this idea by pointing out that even if we were an expert about the machinery of a bat, we could not imagine what it is like to be a bat, what it feels like to be a bat, or more generally what it feels like to be anyone else. Here, we concentrate on subjective visual experience, the sensation that sometimes accompany neural visual processing (Kanai & Tsuchiya, 2012).

In the last 20 years, the search for the neural correlates of consciousness has been very active. Leaving aside concepts and theories, most studies on visual consciousness adopted a pragmatic approach (Crick & Koch, 1990), and contrasted neural responses to stimuli that were consciously perceived vs. stimuli that remained unnoticed. But what does such a contrast tell us about visual consciousness? Does it pertain to information processing that could take place in a robot, or does it pertain to the neural basis of subjective experience? Since these questions were most often not explicitly addressed, subjective visual experience remains an underspecified issue in cognitive neuroscience. On the other hand, the nature of subjective perceptual experience is hotly debated from a theoretical point of view. Some philosophers argue that subjective experience is a cognitive illusion (Dennett, 1991, O'Regan and Noe, 2001), a post-hoc cognitive reconstruction rather than an immediate experience (Cohen and Dennett, 2011, Dehaene et al., 2006), whereas others emphasize that subjective experience is central to consciousness and is distinct from cognitive abilities (Block, 2007). At the other end of the spectrum, many visual scientists do actually study consciousness without mentioning it, since any study on explicit visual perception pertains to visual consciousness.

In this paper, we review neural and behavioral studies on visual recognition, whether or not addressing directly and explicitly the issue of subjective experience, and show how the architecture of the visual system constrains conscious perception. In the light of those constraints, we propose that the distinctive feature of initial subjective experience is coarse vividness: initial subjective experience is rich, integrated and meaningful, but does not contain much details. We provide a physiologically plausible model that accounts for this property and that articulates the quality of subjective experience with neural information processing. We then show how two intensely debated issues in consciousness research, namely the interpretation of change blindness studies and of the Sperling experiment, and the dissociation between attention and consciousness, can be parsimoniously interpreted in the framework of the coarse vividness hypothesis.