Brain activity underlying visual perception and attention as inferred from TMS–EEG: A review

Abstract

Probing brain functions by brain stimulation while simultaneously recording brain activity allows addressing major issues in cognitive neuroscience. We review recent studies where electroencephalography (EEG) has been combined with transcranial magnetic stimulation (TMS) in order to investigate possible neuronal substrates of visual perception and attention. TMS–EEG has been used to study both pre-stimulus brain activity patterns that affect upcoming perception, and also the stimulus-evoked and task-related inter-regional interactions within the extended visual-attentional network from which attention and perception emerge. Local processes in visual areas have been probed by directly stimulating occipital cortex while monitoring EEG activity and perception. Interactions within the attention network have been probed by concurrently stimulating frontal or parietal areas. The use of tasks manipulating implicit and explicit memory has revealed in addition a role for attentional processes in memory. Taken together, these studies helped to reveal that visual selection relies on spontaneous intrinsic activity in visual cortex prior to the incoming stimulus, their control by attention, and post-stimulus processes incorporating a re-entrant bias from frontal and parietal areas that depends on the task.

Introduction

Some of the recent methodological developments in cognitive neuroscience offer new tools with which to trace the passage of information through the brain. This potentially enables measuring how information is initially represented at an early stage of neural processing and is filtered, selected, and allowed to pass through several drafts of increasingly deep analysis, until a percept is formed. The various processes leading to a percept can be conceived as falling into two groups: those that relate to local patterns of brain activity within early sensory areas receiving, analysing and/or passing on information, and those describing longer–range interactions within a larger network for a flexible adaptation of these local processes to varying stimuli and task contexts.

In occipital areas, visual-attentional selection may be in part mediated by populations of neurons which are thought of as inherently interacting or competing with one another [1]. These processes can be modulated by a distributed attentional control network including the frontal eye fields (FEF), posterior parietal cortex (PPC) and also more ventral regions in both frontal and parietal lobes [2], [3]. Many of the experiments that have explored the neural mechanisms underlying attentional orienting and selection used a covert attention task (the Posner task), in which a centrally-flashed symbol cues the likely location of an upcoming visual stimulus, which has to be discriminated [4], and where spatial attention facilitates performance. If monkeys are trained to covertly attend to a peripheral spatial location, it can be shown that the spontaneous activity in visual neurons representing the attended location increases, even before the target is presented [5]. It is noteworthy that this activity change occurs before the onset of the stimulus, as this excludes that such responses in visual neurons are driven by the visual target instead of attention. These baseline shifts in spontaneous visual activity may instead support an increase in processing weight allocated to the attended location. Similar shifts have also been demonstrated in the human brain with functional magnetic resonance imaging (fMRI) [6] and may relate to baseline changes that can be seen with EEG.

Some of the TMS–EEG work described here has indeed linked posterior EEG activity at baseline to the sensitivity of occipital cortex to respond to upcoming external stimulation. If visual neurons are in a state of high excitability at the time of visual input, it should then be easier for visual information arriving in visual cortex to be picked up, analysed, and passed on, than if the cortex is not so excitable. TMS here offers an interesting means to measure the state of visual areas at a given time point, because it elicits crude, flash-like visual percepts (‘phosphenes’) from stimulating visual cortex, given careful manipulation of TMS intensity and coil orientation [7]. By occipital TMS and sampling phosphene report, the excitability of visual cortex can therefore be estimated more directly, bypassing some of the subcortical structures that may affect visual input when stimuli are instead presented to the retina. It has been demonstrated that if the underlying cortex is particularly excitable at the time of the TMS pulse – for example if participants are covertly attending to the location in which the phosphene is expected [8] – then it is easier to elicit a phosphene and the phosphene threshold intensity is lower. Interestingly, in parallel with the attentional modulation of visual cortex excitability [8], oscillatory EEG activity over posterior regions, in particular in the alpha-band (8–14 Hz), is also modulated, and it has been suggested that this alpha-activity may play an active role in mediating attentional competition effects between targets and distracter [9]. High alpha power over posterior areas may serve suppression of distractor information at visual field positions opposite to the alpha power change [9], in line with the alpha-inhibition hypothesis [10]. Conversely, low alpha power may serve the selection of target information at corresponding positions. Could alpha-activity then relate to the sensitivity of occipital cortex at baseline to respond to upcoming external stimulation, possibly under the control of higher-order attention areas? One way to explore this further would be to employ brain stimulation to disrupt these baseline shifts and the presumably dependent perceptual processes.

One goal is then to relate behavioural and neural measures to each other, to study how changes in sensory areas and long-range task-related cortico–cortical interactions arise, affect one another and give rise to the formation of a percept. To do this, each of the methods available to cognitive neuroscientists can uniquely address particular questions. In this review, we will focus on recent work which has combined TMS and EEG in order to explore the functional dynamics of perception and attention. Others have recently reviewed the methodological or conceptual bases for using TMS with EEG [11], [12], [13], [14] and the general logic of combining TMS with other methods in cognitive neuroscience [15], [16], [17], [18], [19].