Can nonconceptual content be stored in visual memory?

Abstract

Dartnall claims that visual short-term memory (VSTM) stores nonconceptual content (NCC), in the form of compressed images. In this paper I argue against the claim that NCC can be stored in VSTM. I offer four reasons why NCC cannot be stored in visual memory and why only conceptual information can: (1) NCC lasts for a very short time and does not reach either visual short-term memory or visual long-term memory; (2) the content of visual states is stored in memory only if and when object-centered attention modulates visual processing and this modulation signifies the onset of the conceptualization of that content; (3) only categorical high-level information that characterizes conceptual content and not metric and precise iconic information that characterizes NCC can be stored in visual memory for long periods; and (4) if NCC were stored in visual memory then this would allow recognitional judgments pertaining to NCC—one could recognize the precise shade of a color that one had seen before. However NCC does not allow such recognitional judgments.

Keywords:

• Attention
• Memory
• Nonconceptual Content

Notes

Athanassios Raftopoulos is a Professor in the Department of Psychology at the University of Cyprus.

Notes

[1] I would like to thank an anonymous reviewer for bringing up this point.

[2] To explain why a de re relationship does not involve sortals by bringing us in direct contact with the world one should show how such a relationship is established through a nonconceptual demonstrative reference to worldly objects. For such an account see Raftopoulos (2009).

[3] The account of NCC I have provided does not specify the nature of “content.” The matter is complicated and largely immaterial to my argument in this paper and that is why I will say only a few things. Contents can be Fregean entities involving modes of presentations of objects in perception or thought. Of course classical Fregean modes of presentation consist of concepts and, in order to accommodate within a Fregean context the notion of NCC, one must either extend the Fregean framework to include nonconceptual modes of presentation to make possible a Fregean account of NCC (Raftopoulos, 2009; Raftopoulos & Muller, 2006b), or one may follow Tye (2006) by arguing that the modal definition of NCC paves the way for a Fregean account of NCC since a Fregean would argue that the concepts that enter into the specification of that content by means of conceptual modes of presentation need not be concepts that the subject of the experience possesses. Alternatively, contents may be Russellian or Millian entities involving not modes of presentation but the objects themselves and their properties. If one extends the Fregean framework to include nonconceptual modes of presentation then, to the extent that these modes bring us in direct de-re contact with the world, one has to include the objects themselves, in addition to their modes of presentation, in the specification of the content of perceptual states, adopting thus a neo-Fregean (Recanati, 1993, chapters 2 & 3) Contents can also be categorized depending on whether one thinks that they are individuated on the basis of what transpires within the brain (internalism-narrow contents), or whether the world contributes to their individuation (externalism-wide contents). In the latter case, the references of the referring terms in the complex term expressing the content are parts of the content. If one holds, as I do, the view that perceptual NCC is retrieved directly from a scene through a de-re relationship with the world, one has to espouse the externalist view and consider the contents of perceptual states as wide contents. One could adopt a hybrid “two-component” picture, and hold that representational states have both wide and narrow contents, which together constitute the complete content of the state.

[4] A way to defend the conceptual modulation of the content of cognitively impenetrable states would be to posit the existence of concepts hardwired or embedded in perceptual circuits either from the beginning (nativism) or as a result of the development of visual circuits. The content of these states could be conceptual even though no cognitive signals penetrate perceptual processing. However, if there are concepts hardwired in perceptual mechanisms that are otherwise cognitively impenetrable, then these “concepts” do not play the role concepts are usually thought to play in cognition. First, they are not personal contents since one is not aware of them when they are employed in early vision. Second, they can be used only in the processes of early vision and they are not available for cognitive tasks (Pylyshyn, 2007; Raftopoulos, 2001a, 2001b, 2009). Third, they do not allow re-identification across times and contexts of the objects formed during early vision (Campbell, 2006; Heck, 2007; Kelly, 2001; Raftopoulos, 2009). Fourth, they do not satisfy Evans’ (1982) generality constraint (Heck, 2007). For these reasons, Pylyshyn (2007, p. 52) calls them subpersonal concepts since, being inside encapsulated modules, they do not enter into general reasoning. Thus, if one uses “concept” with its usual significance, this innateness does not entail that early visual states have conceptual content.

[5] Attention acting as a hand that grasps the proto-objects is best understood as a metaphor for the fact, explained by the theory of attention as a biased competition that I discuss later, that certain proto-objects win the competition against other proto-objects and they are selected (picked by attention's hand) for further processing.

[6] The proposal that consciousness results from global recurrent processing involving higher areas in the brain, such as frontal and prefrontal areas and mnemonic circuits, accords with Dehaene et al.'s (1998) work that puts forth the hypothesis of the “global neuronal workspace model.” According to this hypothesis, the step to conscious experience consists in the entry of processed visual stimuli into a global brain state that relates distant areas, including parietal, prefrontal, frontal areas, and anterior cingulated nodes.

[7] Here, I rely on comments by Ron Rensink in private communications on February 22, 2009 and March 8, 2009. I would like to extend my gratitude to Ron for helping me understand his theory on visual processing and for some very interesting and helpful discussions on the relation between memory and attention.

Additional information

Notes on contributors

Athanassios Raftopoulos

Athanassios Raftopoulos is a Professor in the Department of Psychology at the University of Cyprus.