RTG 2175 Perception in Context and its neural Basis

Breadcrumb Navigation


New TP Geyer/Müller "Attention templates in visual working memory"

PIs: Thomas Geyer, Hermann Müller (psychology)
Project partner: Paul Taylor (medicine)

In this new collaboration between psychology and medicine, we plan to investigate how differences in search template precision are expressed in electro- and neurophysiological variables using EEG and fMRI. Furthermore, we seek to understand how precision affects the initial acquisition and subsequent updating of template representations, particularly how specific areas of the fronto-parietal attention network are causally involved in these mechanisms using neurostimulation techniques.

Current theories of attention assume that human observers maintain a template representation of the task-critical stimulus in visual working memory (vWM) to bias perceptual processing toward relevant features. This is typically shown experimentally by presenting a feature cue prior to a search array, allowing participants to establish a precise template representation for the subsequent search.

In everyday life, however, clear cues are rarely present to inform us about which features are important for our search and when the stimuli will appear – thus, we cannot benefit from high template precision to expedite responses. While we have some crude idea about the target of our search (e.g., find an edible berry amongst inedible leaves and/or toxic berries), we cannot anticipate the exact target appearance (i.e., is the edible berry reddish or reddish-blueish) on the basis of this imprecise template representation. Imprecise search templates will ultimately (1) draw attention to a relevant subset of items (including irrelevant “distractors” as well), and (2) need to be updated once attention has been deployed to the target item, in which case the target’s identifying features can be extracted more accurately.


Event-related potentials that index the maintenance of objects in vWM like (the amplitude and latency of) the contralateral delay activity (CDA) will be compared under precise and imprecise search template conditions. The former will be established with a feature cue indicating the exact identity of the upcoming target, facilitating top-down guidance in the subsequent search array. In contrast, participants will only be able to discern the exact target identity when it’s found in the search array. This will be accomplished by flashing the search display at brief intervals interrupted by longer blank screens to determine the discrete point in time when participants start to refine their – still imprecise – search template representations and detect the target item (see Fig. 1). This will be followed by the medical portion of the project where we will test how BOLD activations overlap and differ during the acquisition and updating of search templates using the new MCN-LMU MRI scanner (hypothesized regions are depicted in Fig. 1). Using, in the first instance, an online event-related 10-Hz TMS protocol, we will then test the causal contributions of specific brain structures associated with the different processing stages.


Figure 1. Left: schematic illustration of the (interrupted) search paradigm used to investigate predictive coding of the target’s perceptual features across subsequent looks of the visual search array (in the illustration, the target is the letter “T” oriented 90° in anti-clockwise direction). Top-right: participants’ behavioral search performance (i.e., percentage of fast responses) as a function of individual looks or epochs. Bottom-right. Likely neural generators of the working-memory guided interrupted search (SFG: superior frontal gyrus; rTPJ: right temporo-parietal junction).