B.3.6.1.3. Integration between endogenous and exogenous processes

Endogenous and exogenous attentional processes are not mutually exclusive, and they can act concurrently to orient attention. For instance, Müller, Reimann & Krummenacher (2003) showed that knowing in advance the colour of the salient target to come could enhance the performances in a visual search paradigm. These results were confirmed in a trial cueing mode precluding perceptual priming influences (Müller & Krummenacher, 2006). These data showed that endogenous featural knowledge could be combined with exogenous influences to orient attention. More precisely, they showed that knowing the dimension or feature in which the target had to be salient helped attentional selection of this target. In our lab, we obtained some prior evidence that endogenous knowledge about a feature that was neither salient nor defining the target could also improve attentional selection of an otherwise salient target. This suggested that two distinct signals –endogenous and exogenous– could have additive and independent effects on attentional orienting.

The data reviewed in previous sections evidenced that top-down (relevance) and bottom-up (salience) effects could theoretically target the same perceptual sub-systems and have concurrent effects thereupon. For instance, to continue with colour, endogenously attending a colour could modulate response in the V4 visual area (Anlle-Vento et al., 1998; Bichot et al., 2005; Schoenfeld et al., 2007), which is highly involved in processing colour (Bartles & Zeki, 2000). Similarly, the activity of this area could also be modulate by colour-induced salience (Burrows & Moore, 2009; Beck & Kastner, 2005). Therefore, neuronal populations responding to a red stimulus in their receptive field, could have their activity enhanced by both exogenous effects from salience (by homogeneous green item surround) and endogenous effects from a template in working memory (by prior cueing of the target's red colour). About their results, Beck and Kastner (2005) stated that they:

‘"suggest that both top-down mechanisms related to spatially selective attention and bottom-up mechanisms related to stimulus context operate by resolving competitive interactions at intermediate processing stages in visual cortex" (p. 1114).’

Saying that, these authors considered spatial endogenous attention. However, endogenous featural attention was postulated to act similarly to spatial attention by Desimone and Duncan (1995; see also Beck & Kastner, 2009). On the basis of both these theoretical considerations and empirical evidence, we thus derived a tentative hypothesis about this endogenous-exogenous integration.

Let us consider that endogenous and exogenous attentional influences affected competition in the same cognitive systems, for instance the one system specialized in processing colour, or movement, or so on. This was supported by the data reviewed so far. If this is true, both influences could take place either at the same or at a different locus, depending on whether they concerned the same or a different dimension, respectively. Working on this assumption, one could expect that the strength of their interaction could vary according to the similarity of the dimension concerned by each factor (endogenous and exogenous). Indeed, it is possible that, if both salience and relevance concerned the colour, they had interactive effects. For instance, if the target was salient in colour (e.g. a red amongst green items), and if this target's colour was also cued before the visual search, one might possibly expect that both signals would genuinely integrate, as they might for instance coactivate the same neuronal population in the colour processing perceptual system. On the contrary, one might expect no such integration between these signals if they concerned different features. Indeed, if salience concerned "colour" and relevance "motion", for instance, then each signal could tap a different perceptual sub-system, and integration should then occur only higher in perceptual processing. This variable integration hypothesis could be derived from the biased competition models, although this is not a necessary consequence. Furthermore, the precise neuronal mechanisms involved in this hypothetical integration are only tentatively outlined. They might be very different and should be examined directly by future research. The core objective of this article, on the basis of this hypothesis, was only to suggest: a) that endogenous and exogenous processes of featural attention could genuinely integrate and b) that this integration might take place within the perceptual system, as suggested by the biased competition hypothesis. These results might have important consequences for our understanding of visual selective attention.

The present research aimed at testing this hypothesis of contingent integration, in order to support the biased competition models. To that end, a visual search task was proposed, in which endogenous and exogenous factors were manipulated independently (see Figure 8). While an item (target or distractor) could be salient in one of two features (colour or size), the cue could only indicate (or not) the colour of the target. The interaction between salience, relevance and featural similarity was tested by a factorial ANOVA on correct RTs. More crucially, a genuine integration between endogenous and exogenous factors was also tested for both similarity conditions, using the Miller's (1982) inequality (see Data analysis, section B.3.6.2.5).