Who controls the attentional enhancement of targets, the blocking of distracters, and the resolution of conflict?

One of the things that distinguishes humans from lower animals is their ability to select task-relevant information in cluttered sensory environments. This ability goes beyond merely selecting the information to which one has to respond; it also involves the ability to deal with the conflict that emerges when multiple stimuli elicit incompatible responses. This response conflict results from insufficient selection between the different stimuli. Of course, at some point, this conflict has be resolved; otherwise, no response would be given. The dominant view is that the conflict is resolved by a stronger stimulus selection (i.e., enhancing task-relevant information and blocking distractors). One of the central questions in cognitive neuroscience is how the brain achieves this selective gating of sensory information in interaction with the detection of response conflict.

A dominant view is that neuronal oscillations play a central role in this gating of sensory information. This starts from the common observation of alpha band (8-14 Hz) oscillations over posterior (occipito-parietal) areas and alpha- and beta band (15-25 Hz) oscillations over sensorimotor areas. With respect to the functional role of these oscillations, the dominant view involves that high amplitude neuronal oscillations block the sensory input (visual for the posterior and somatosensory for the sensorimotor areas) whereas low amplitude oscillations allow the sensory input to be transferred to the downstream areas that are responsible for cognitive control and motor output.

These observations have led to the view that low amplitude neuronal oscillations allow the sensory input to be transferred to their downstream targets whereas high amplitude neural oscillations block the sensory input. The question now is, via which mechanism this gating of sensory input is controlled. The dominant view here is that frontal cortical areas are responsible for this, and this view is consistent with the fact that fMRI studies of attentional control show a robust involvement of these frontal cortical areas. Compared to fMRI studies, there is only limited evidence from electrophysiological studies showing the involvement of frontal cortical areas in the modulation of neural oscillations over sensory input areas. Two recent magnetoencephalography studies are exceptions to this rule (Baldauf & Desimone, 2014; Sacchet et al, 2015), and both found evidence for oscillatory coupling between the right inferior frontal cortex (rIFG) and different sensory areas (S1 in Sacchet et al, 2015; fusiform face area and parahippocampal place area in Baldauf & Desimone, 2014). An important limitation of these studies, however, is that they do not distinguish between target enhancement and conflict resolution. Making this distinction is a central objective of our study. This central objective can be split in two questions:

  1. Do attention-induced amplitude modulations over sensory areas reflect target enhancement, distracter suppression, or both?
  2. Which brain areas control stimulus selection (involving target enhancement and/or distracter suppression) in sensory areas, and how do these sensory and control areas communicate?

Figure 1. Snapshot of an example stimulus stream

To answer these question, we make use of a novel experimental paradigm. This paradigm involves trials in which the participant fixates the center of the screen while stimulus streams are presented in the left and the right visual field. The participant responds using two buttons, one for his left and one for his right hand. One of the two buttons has to be pressed depending on the content of the attended stimulus streams, as explained in the following. By means of a cue, one of these two streams will be indicated as the task-relevant one, and the other one will be the distracter. The streams are continuously present but their content varies over time. There are two stimulus categories; for concreteness, we assume them to be letters and digits. Each of the two buttons is associated with one stimulus category (letters or digits), and the participant has to press this button when the fixation dot increases size (the so-called go-signal). A snapshot of an example stimulus stream is shown in Figure 1. Over the course of time, the stimulus streams change, with letters being replaced by digits or other letters, and vice versa for digits that are being replaced. At any point in time, one of the task-relevant stimulus streams indicates the hand (left or right) with which the participant must press the corresponding button after the go-signal. The relevant stimulus stream is indicated by the colour of the fixation dot, for example, with yellow denoting left and blue denoting right. The colour of the fixation dot is the so-called cue.

Figure 2. Snapshot of an example varying-contrast stimulus

For answering our research questions, it is required that target enhancement and distracter suppression are manipulated independently, and this not the case for the trial type just described. In fact, because both target and distracter are continuously present they are perfectly anti-correlated. Therefore, we now change the stimulus streams such that target and distractor become uncorrelated. The resulting new stimulus type is called a varying-contrast stimulus, and a snapshot is shown in Figure 2, which shows a moment at which only the right stream has a low contrast. The contrast of both streams continuously varies over time, and this can be achieved by degrading the letters and digits (arrays of square pixels whose grey-values can be adjusted to achieve the appropriate contrast; see Fig. 2). Over time, for each of the streams, the contrast is modulated rhythmically between maximum and zero contrast. The rhythmic contrast modulations of the two streams have different frequencies (e.g., 1 Hz for the one and 1.5 Hz for the other stream), which results in a zero temporal correlation between target enhancement and distracter suppression (if the time window is sufficiently long). This is a form of frequency tagging, which has the important advantage that a frequency domain analysis of the electrophysiological data allows for a separation of the neurophysiological correlates of target enhancement and distracter suppression (involving both cognitive control processes and their consequences over sensory areas). This separation can be performed effectively by means of a regression analysis. Importantly, within each of the two streams, the stimuli will vary across the two stimulus categories, each of which is associated with one response side. This has several advantages: (1) it requires a continuous engagement over the course of a trial, and (2) it allows to investigate stimulus selection during both stable and switching motor preparation. The latter is likely to be a crucial variable for cognitive control, whose neuronal substrate is closely related to the motor system.

This project will be supervised by Eric Maris.