B.3.6.1.1. Salience

Salience is an important determinant of visual selective attention (Nothdurft, 2006). It is induced by the feature contrasts in the visual field (Nothdurft, 1993; Wolfe & Horowitz, 2004). Of particular importance for salience are the homogeneity between distractors, and the target-distractor similarity (Duncan & Humphreys, 1989), that is, the overall featural contrast. The underpinnings of salience have been found in the visual system. From the primary visual cortex (V1), the perceptual system seems to segregate in several sub-systems relatively specialized in computing various aspects of the visual scene, such as colour, motion, and so on (Felleman & Van Essen, 1991). This segregation was used in many theories of visual selective attention (e.g. Treisman & Gormican, 1988; Koch & Ullman, 1985; Wolfe, 1994, Michael et al., 2006). One interesting characteristic of these visual maps is that they do not code the features per se, but rather the feature contrasts, probably on the basis of a competitive, mutually inhibitory, organisation (Knierim & Van Essen, 1992; see also Jones, Wang, & Silito, 2002; Kastner & Beck 2005; Kastner, Nothdurft, & Pikarev, 1997; Nothdurft, Gallant, & Van Essen, 1999; and Burrows & Moore, 2009, for similar effects in V4).

In these experiments, the presence of surrounding distractors, outside the classical receptive field, decreased the neuronal response to a preferred target This supported the hypothesis of a competition between items (Duncan, 2006). Crucially, these competitive effects decreased when the similarity between the preferred stimulus and its neighbours increased (i.e. when the contrast decreased). This is directly reminiscent of the well known psychophysical effects of target-distractors and distractors-distractors similarity on visual search, reported by Duncan and Humphreys (1989). Therefore, these competitive mechanisms are probably the basis of salience effects (Nothdurft, 1993, 2006). They suggested that salience effects could emerge within the perceptual system, and bias the competition taking place therein between visual objects.

This hypothesis is in strong opposition to the salience map models (e.g. Itti & Koch, 2001; Michael et al., 2006; Wolfe, 1994). These models, for their part, assumed an independent map representing a weighted sum of these saliences. One audacious postulate of these models was that there would remain no featural information anymore in this salience map, which represented salience, merely and strictly (Koch & Ullman, 1985). Then, the activations on this map would only be subsequently responsible for the attentional enhancement of the perceptual processes. VanRullen (2003) criticized this counter-intuitive proposition. On the other, the biased competition hypothesis considered that exogenous attentional effects emerged within the perceptual system and thus directly (rather that indirectly) modulated the perceptual processes.