The temporal flexibility of attentional selection in the visual cortex Jens-Max Hopf, Mircea Ariel Schoenfeld and Hans-Jochen Heinze Visual attention operates by biasing competitive interactions between neural representations, favoring relevant over irrelevant visual inputs. Attention can enhance the processing of relevant information using location-based, feature-based or object-based selection mechanisms. Studies using event- related potential and event-related magnetic field recordings, together with functional magnetic resonance imaging, show that the temporal sequencing of these different selection mechanisms is flexible. Depending on the specific processing demands of the experimental task, location-based, feature- based or object-based selection might be given temporal priority on a time scale of tens of milliseconds. Addresses Department of Neurology II, Otto-von-Guericke University and Leibniz Institute for Neurobiology, Leipziger Strasse 44, D-39120 Magdeburg, Germany Corresponding authors: Heinze, Hans-Jochen (hans-jochen.heinze@medizin.uni-magdeburg) and Hopf, Jens-Max (jens-max.hopf@medizin.uni-magdeburg.de) Current Opinion in Neurobiology 2005, 15:1–5 This review comes from a themed issue on Cognitive neuroscience Edited by Angela D Friederici and Leslie G Ungerleider 0959-4388/$ – see front matter # 2005 Elsevier Ltd. All rights reserved. DOI 10.1016/j.conb.2005.03.008 Introduction When we are faced with an overwhelming set of sensory inputs, attention mechanisms select the information that is relevant for our behavioral goals. According to the influential ‘biased competition theory’ [1–4], selectivity is accomplished in the neural system by enhancing the neural representations of relevant input sources, giving them a competitive advantage over representations of irrelevant input sources. This biasing of sensory repre- sentations is mediated by top-down control from fronto- parietal circuits [5], and also by automatic bottom-up processes such as perceptual popout. Sites of competitive enhancement or attenuation have been identified at almost every level in the human visual processing hier- archy, including the lateral geniculate nucleus and pulvi- nar [6], striate cortex [7–11,12  ] and several areas of extrastriate cortex [13–18,19  ,20  ]. Importantly, it has been demonstrated that location-based, feature-based and object-based mechanisms can be employed to gain competitive advantage [21]. Researchers have debated which one of these selection mechanisms is more basic [22–27]. By virtue of having excellent temporal resolu- tion, event-related potential (ERP) or event-related mag- netic field (ERMF) recordings in human observers have provided new insights into the fine spatiotemporal struc- ture of these mechanisms. In accord with psychophysical observations [25], ERP and ERMF studies have shown that location-, feature- and object-based attentional pro- cessing is coordinated in the visual cortex on a tight spatiotemporal scale, in a flexible manner that depends on ongoing processing needs. The temporal priority of location selection When selection is based on spatial location, electromag- netic recordings have shown that attended-location sti- muli elicit larger sensory responses than ignored-location stimuli from 80–300 ms after stimulus onset [28,29]. Specifically, the selection of spatial locations leads to an enhancement of the P1 component (80–100 ms) and the N1 component (130–200 ms) [30,31]. By contrast, selec- tion based on nonspatial features such as color or orienta- tion does not modulate these early sensory responses, but instead leads to longer-latency ‘selection negativities’ or ‘selection positivities’ [30,32–35]. Furthermore, modula- tions of the P1 and N1 components were found to reflect different aspects of attentional selection, with the initial P1 modulation reflecting location selection in particular, and the subsequent N1 modulation reflecting stimulus discrimination within the focus of attention [36–38]. An additional ERP–ERMF index of location-based atten- tional selection — the N2pc component — is observed in visual search tasks, in which a target is present at an unknown location within an array of distractors. The N2pc is a negative-going voltage deflection in the N2 latency range (onset = 170–200 ms) over the posterior cortical areas contralateral to the location of the target item (N2pc = N2-posterior-contralateral). The N2pc component reflects the focusing of attention onto the target [39–43], and it is usually the first detectable ERP– ERMF attention effect when the location of the target is not known in advance. Source localization based on ERMF data revealed that initial parts of the N2pc origi- nate from current sources in the posterior parietal cortex (180–200 ms) followed by later parts arising from current sources in occipito-temporal cortex areas (after 200 ms) [43]. Woodman and Luck [41] further demonstrated that the N2pc component shifts rapidly between the hemi- spheres as subjects shift attention from distractor to distractor while they search for the target, thereby pro- CONEUR 254 www.sciencedirect.com Current Opinion in Neurobiology 2005, 15:1–5