Purchase this article with an account.
Kyoung whan Choe, Randolph Blake, Sang-Hun Lee; Decomposition of stimulus representations and decision-bias signatures in population activity of human primary visual cortex. Journal of Vision 2013;13(9):1264. doi: 10.1167/13.9.1264.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The precise role of primary sensory cortex in forming perceptual choices remains elusive. In some studies trial-to-trial fluctuations in responsiveness within primary visual cortex (V1) are tightly correlated with observers’ perceptual responses to repeated presentations of otherwise identical stimuli (Ress and Heeger, 2003; Palmer et al., 2007); in other studies no such correlation is observed (Grunewald et al., 2002; Nienborg and Cumming, 2006). At the same time, non-sensory factors such as task structure (Jack et al., 2006; Cardoso et al., 2012) or decision bias (Nienborg and Cumming, 2007) do appear to impact neural activity within early visual cortex. To elucidate V1’s contributions to perceptual decisions, we measured fMRI responses from V1 while 19 human observers performed a difficult visual discrimination task (performance: 73.7% ± 5.7% correct) and assessed simultaneously whether and how population activity in V1 represents stimuli presented and signifies choices and decision biases exhibited by observers. Over trials one of three briefly presented (0.3s), different-sized rings (2.78°, 2.84°, and 2.90°, randomly presented) was categorized by observers as ‘small’ or ‘large’. A large fraction of fMRI responses in V1 fluctuated in unison, leading us to decompose raw fMRI response profiles as a function of eccentricity (RRs) into ‘untuned responses’ (URs, average of RRs across eccentricities) and ‘tuned responses’ (TRs = RRs-URs). TRs readily resolved threshold-level differences among stimuli with high fidelity (stimulus probability: 0.57 ± 0.06; mean ± SD) but failed to predict perceptual choices made by observers. URs, on the other hand, correlated significantly with the inter-observer variability in decision bias (Pearson’s r=0.52, p=0.02). These results suggest that population responses in V1 encode external stimuli with high fidelity but their trial-to-trial variability is unlikely to signify trial-to-trial variability in actual choice selection. The relationship between untuned V1 activity and decision bias could presage an expanded conceptualization of global fluctuations in early visual cortex, perhaps reflecting sensory gating via neuromodulatory feedback mechanisms.
Meeting abstract presented at VSS 2013
This PDF is available to Subscribers Only