Introduction: Numerosity, the set size of a group of items, is processed by association cortex, but certain aspects mirror properties of primary senses (Dehaene, 1997, OUP; Burr and Ross, 2008, Current Biology). The parietal cortex contains numerosity-tuned neurons. We recently demonstrated that numerosity-tuned responses are organized topographically, but that population numerosity preferences vary with stimulus features (Harvey et al., 2013, Science). How is numerosity selectivity related to the low-level visual features from which numerosity is computed? Methods: We used high-field (7T) fMRI and custom-built analyses to capture variations in the time course of numerosity-selective voxels resulting from differences in preferred numerosity and tuning width, following a population receptive field (pRF) design (Dumoulin and Wandell, 2008, Neuroimage). Several control conditions were tested to check that low-level stimulus features that co-vary with numerosity (such as pattern edge length, pattern surface area, pattern density, and individual item size) did not underlie response selectivity. We quantify low-level features in these stimuli to investigate their effects on numerosity selectivity. Results: Numerosity explained population responses far better than any low-level feature. However, numerosity preferences can be affected by stimulus features in some subjects when features vary incongruently with numerosity. These interference effects are found particularly when display luminance contrast and item size decrease strongly when numerosity increases. Even in this situation, populations remain tuned to numerosity and the topographic representation of numerosity remains in the same direction. Conclusions: Numerosity, not low-level features, is the primary stimulus dimension underlying numerosity-selective responses and cortical organization. However, numerosity must be derived from early visual feature selectivity. Interactions between numerosity tuning and stimulus features suggest how numerosity may be computed (Dakin et al, 2011, PNAS). Alternatively, effects of both numerosity and object size on population tuning suggest that these different magnitudes may be processed together (Walsh, 2003, TICS).
Meeting abstract presented at VSS 2014