Perceptual and Attentional Mechanisms within the Human Lateral Occipital (LO) Region: An rTMS Approach.

Any natural visual environment contains a huge collection of objects, which impact on our perception and compete for drawing our interest and therefore for being preferentially noticed. By effectively selecting a relevant fraction of the incoming information for further in-depth processing, visual selective attention (VSA) optimizes vision in order to overcome the intrinsically limited computational capacity of the visual system. Single-unit recording studies have demonstrated that multiple stimuli simultaneously impinging onto the receptive field (RF) of a given neuron compete for controlling its firing by interacting with each other through mutual inhibition (Reynold & Chelazzi, 2004; Chelazzi et al., 2011; see also Biased competition model by Moran & Desimone, 1985). Thus, neural responses to stimulus pairs in the RF approximate a weighted average of the responses elicited by individual stimuli (for further details, see Reynold & Chelazzi, 2004; Chelazzi et al., 2011; see also Normalization model by Reynolds et al., 1999). The crucial question to ask is how the competition is resolved. Neurophysiological studies have shown that when two stimuli are simultaneously presented within the same receptive field (RF), neuronal responses in the absence of attentional control are largely determined by the strongest or most salient stimulus, e.g. the one presented at higher luminance contrast, which stands conspicuously against the background (Reynolds & Desimone, 2003). This reflects a bottom-up biasing of the competition on the basis of stimulus saliency. Crucially, top-down attentional control can resolve the competition between stimuli in favor of the most behaviorally relevant stimulus (target) by specifying its properties. In other words, attention can switch control of the neuronal response to the stimulus of interest, independently of its saliency, so that the target will determine the response of that neuron; in other words, the response of a given neuron to a pair of stimuli impinging on its RF will equate the neuronal response to the target stimulus, when presented alone. As a consequence, the neuronal representation of the target is enhanced within visual areas at the expense of the visual representation of the distractor (Corbetta et al., 1990; Treue & Trujillo, 1999; Luck et al., 2000, for reviews, see Chelazzi et al., 2011; Carrasco, 2011; Roe et al., 2012). Importantly, there is a wide range of observations which describe the impact of attention on sensory representations along the visual pathway. Crucially, attentional biasing of the neuronal activity within visual cortices is not uniform, but rather results in different forms of neuronal modulation (see e.g. Treue & Martinez Trujillo, 1999; Fries et al., 2001, 2008; Martinez-Trujillo & Treue, 2002; Carrasco et al., 2000; Carrasco, 2006). The traditional view of VSA maintains that attentional control is organized in a master-slave hierarchical manner: Modulatory top-down signals from a distributed frontoparietal attentional network (e.g. Moore, Armstrong, 2003; Wardak et al., 2004; Silvanto et al., 2006) - the master - impact on sensory (visual) cortical areas - the slave. In other words, lower-order sensory areas execute visual representations commanded via feedback projections from higher-order centers. Recent research has greatly challenged this conventional view, leading to the new and striking hypothesis that master centers do not have an exclusive role in attentional control, but rather the slave ventral (and dorsal) visual pathway areas might capitalize on their internal microcircuitry to directly instantiate attentional mechanisms even in the absence of control from master centers (e.g. Reynolds & Heeger, 2009; Baluch & Itti, 2011). Interestingly, a behavioral assessment following circumscribed lesions of macaque areas V4 and TEO showed a strong impairment in the animal ability to select a stimulus based on its behavioral relevance while discarding other, perceptually more conspicuous stimuli; in other word, after lesions to those areas, the behavior of the animal was at the mercy of stimulus salience (De Weerd et al., 1999; see also Gallant et al., 2000 for analogous findings in humans). These areas along the ventral pathway have therefore been claimed as essential for the instantiation of attentional mechanisms, and in particular mechanisms for the efficient filtering of non-relevant distractors (De Weerd et al., 1999; Chelazzi et al., 2011). The aim of the present study is to extend the current understanding of the brain mechanisms underlying VSA, by directly testing their possible residence within the human object-recognition pathway itself. An excellent human slave candidate to test this possibility is represented by the lateral occipital cortex (LO), a mid-tier area of the ventral stream, which is a key node for shape-object perception (Malach et al., 1995). Specifically, by applying TMS stimulation over human LO (or a control site), we examined the role of LO during a VSA task, in order to directly test its role in the attentional filtering of distracting information. Crucially, we manipulated the timing of TMS application in two related experiments, in order to disentangle the contribution of LO to perceptual and attentional operations. As a result, we observed TMS modulation of activity within LO area during the attentional processing of our VSA task. By using early TMS (before stimulus display onset) and late TMS (during stimulus display onset) application over LO cortex, we obtained more general perceptual enhancement and more specific improvement of attentional filtering, respectively. We can therefore conclude that human slave LO area contains internal attentional microcircuits necessary for attentional target selection and distractor filtering.