Purchase this article with an account.
Piers Howe, Margaret Livingstone, Istvan Morocz, Todd Horowitz, Jeremy Wolfe; A Neurophysiological model of multiple object tracking derived from fMRI. Journal of Vision 2008;8(6):220. doi: https://doi.org/10.1167/8.6.220.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The multiple object tracking (MOT) paradigm is a powerful tool for studying dynamic attention and the nature of objects. However, surprisingly little is known about the neural systems underlying MOT performance. Here we present an improved method for identifying core brain areas involved in MOT, and introduce a novel analysis of effective connectivity among these areas. Previous fMRI studies of MOT employed a “passive viewing” baseline. This baseline is problematic, because it does not allow control of attentional load relative to tracking. We added an additional stationary baseline. Stimuli comprised eight identical disks, four moving and four stationary. Observers (N = 13) attended either to two moving disks (tracking), two stationary disks (stationary) or no disks (passive) while maintaining central fixation. Subtracting the stationary activity from tracking activity yielded a map of areas preferentially activated by MOT. This map comprised FEF, SPL, anterior IPS and MT+. Consistent with previous studies, using the passive baseline would have added posterior IPS and lateral postcentral sulcus. These last two areas are activated by attentional load, but not by tracking per se. Focusing on the first four areas, we employed a novel method to derive effective connectivity from their BOLD signal timecourses. We first determined the feed-forward connections between these areas by using prior anatomical information to constrain the Total Conditioning algorithm (Pellet & Elisseef 2007). This method allowed us to derive causation from partial correlations, while making minimal assumptions. We then reversed the procedure to determine the feed-back connections. This analysis leads to an improved understanding of the relationship between neural activity and performance. For example, we found a surprising asymmetry between ipsilateral and contralateral connectivity patterns, which could explain why behavioral studies have found tracking in two hemifields to be superior to tracking in a single hemifield.
This PDF is available to Subscribers Only