Attention field models capture biases in perceived position

Attention is the mechanism through which we are able to select relevant information from our visual environment. Attention field models describe the effects of attention on neural processing and predict that receptive fields are attracted towards the attended location. We have recently demonstrated this attraction using fMRI and found that the amount of attraction varies across the visual hierarchy (Klein et al, Neuron, 2014). Here, we apply an attention field model to human perception: we predict that receptive field attraction results in a bias in perceived position, which depends on the size of the underlying receptive fields.

Participants were presented with two pairs of Gabors and judged which of the two pairs was spaced closest together. Attention was directed to one of the pairs using exogenous cues. In two experiments we varied (1) the eccentric position and (2) the spatial frequency of the Gabors. As receptive field size increasing with increasing eccentricity and decreasing spatial frequency, we expect a larger bias in the perceived position for higher eccentricities on the hand and lower spatial frequencies on the other.

Our results show that the positional bias increased with eccentricity, but did not vary with spatial frequency. We also found that the increase of the positional bias with eccentricity closely matches the predictions derived from an attention field model. We conducted control experiments to exclude possible alternative explanations and to control for confounds.

In conclusion, attention field models can account for some positional biases, but do not account for the spatial frequency manipulation. We speculate that the attention field as elicited by our cues operates in visual space but not in the frequency domain. Our results suggest that attention field models provide a useful framework to understand effects of attention on human perception.