Academic journal article Perception and Psychophysics

Attentional Templates Regulate Competitive Interactions among Attended Visual Objects

Academic journal article Perception and Psychophysics

Attentional Templates Regulate Competitive Interactions among Attended Visual Objects

Article excerpt

In two experiments, we examined the mechanisms responsible for creating a zone of interference surrounding an attended visual object (see, e.g., Mounts & Gavett, 2004). In Experiment 1, the similarity between attended stimuli and noise items was manipulated in order to contrast an account based on competitive interactions between attended items with an account based on inefficient filtering of unattended stimuli. Consistent with the competitive interaction account, the data revealed that similarity between attended items increased the strength of localized interference, whereas similarity of noise items to the attended stimuli did not Experiment 2 showed that the interference observed between attended items was determined by their match to attentional templates.

The mechanisms of visual selective attention have been variously described as space based (e.g., Eriksen & Hofiman, 1973; Posner, 1980), object based (e.g., Duncan, 1984; EgIy, Driver, & Rafal, 1994), and feature based (e.g., J. Lu & Itti, 2005; Saenz, Buracas, & Boynton, 2002; Treue & Martínez Trujillo, 1999). Recently, these distinct models have begun to be subsumed under the conceptualization of attention as a competition among objects for representation within the visual system (Cave, 1999; Deco & Rolls, 2004; Desimone & Duncan, 1995; Luck, Girelli, McDermott, & Ford, 1997; Vecera, 2000). The competitive interaction framework proposes that stimulus objects compete to be represented by visually selective neurons within the extrastriate cortex. The need for competition is imposed by the neurophysiology of the visual cortex. Receptive fields (RFs) are small in Vl, the earliest cortical visual area, but grow larger through the extrastriate cortex, achieving sizes large enough to encompass two or more objects simultaneously. The result is that a single neuron's response train can conflate the properties of multiple objects (Luck & Beach, 1998; Reynolds & Desimone, 1999; Treisman, 1996). One solution for this problem is to employ a form of noise exclusion (cf. Dosher & Lu, 2000; Z.-L. Lu & Dosher, 1998,2004), filtering signals as they propagate from early to later visual areas, so that the input reaching a given high-level cell arises from only a single stimulus object The neuron thus becomes temporarily dedicated to the representation of a single object (Moran & Desimone, 1985; Reynolds & Desimone, 2003).

Gating of information to higher level cells is controlled by interactions between neurons at multiple levels of representation. To account for increases in RF size across levels of the visual cortex, many competitive interaction models posit a pyramidal processing architecture wherein projections from pools of adjacent neurons converge from one level of representation to the next (e.g., Cave, 1999; Deco & Rolls, 2004; Tsotsos et al., 1995). Competition between objects, driven bottom up by stimulus salience and top down by attentional set (Desimone & Duncan, 1995; Mounts, 2005; Mounts & Gavett, 2004), is instantiated by lateral inhibitory interactions between neurons within each level of the representational hierarchy. During exposure to a visual scene, then, signals from stimulus objects are initially fed forward from lower to higher levels of representation. Following this initial wave of widespread activation, however, competitive processes occurring within each level of representation, and integrated by forward and back projections between levels, begin to prune the signals propagating from lower to higher levels until eventually a single representation (e.g., Hamker, 2004; Tsotsos et al., 1995), or at best a small number of representations (e.g., Bundesen, Habekost, & Kyllingsbaek, 2005), emerges at the highest level of coding.1

Because of the changes in RF size that occur through the processing hierarchy, the strength of the competitive interactions between stimuli is inversely related to their spatial separation. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.