Abstract
The visual cortex is hierarchically organised so that local features are processed in early areas with increasingly complex operations occurring with each subsequent level. Many of these early- to intermediate-stages are represented retinotopically (Wandell, Brewer, & Dougherty, 2005). Within the regions (i.e. V1) there are often populations of neurons which cluster together in terms of their tuning. When the system has to compare information from different populations of neurons (either those within a cortical region or those between regions) this may increase the chance of error affecting the precision of perceptual judgements. The current study investigated the precision of alignment discrimination between forms thought to be processed in separate retinotopic maps, as well as between forms thought to be processed by separate populations of neurons within a map. This precision was compared to that found when alignment judgements were made between forms processed in the same retinotopic map and processed by the same population of neurons, respectively. Our inter-map test used locally detected luminance-defined dots and the centres of radial frequency patterns (which require global processing) while the intra-map test used ‘low’ and ‘high’ spatial frequency Gabor patches. We found there was indeed extra positional error in conditions where an inter-map judgement had to be made. This error could not be accounted for by summing the positional error associated with each individual form within the stimulus set. However, extra error was not found when observers made an intra-map judgement. We conclude that a precision cost may be introduced to alignment judgements when position information is passed between - but not within - cortical regions.