Binding by Synchronisation: a Task-Dependence Hypothesis

It is argued that binding needs to be task dependent, and cannot usefully be driven by properties of the stimulus alone. However, task dependent binding can only take place after the patterns in a stimulus have been identified. Thus pattern recognition needs to be done prior to binding. Synchronisation may be a consequence of pattern recognition and can be used to localise the pattern and tag its attributes at different levels of information processing. -----------------------------------------It is frequently proposed that the function of synchronisation is to bind features belonging to a same stimulus. Let us assume that this is the case. The working hypotheses in early papers on synchronisation were that binding is driven by properties of the stimulus and is mediated by lateral connections within a cortical area (or between areas). However, these hypotheses suffer from several weaknesses: i) The data most frequently cited as evidence that lateral connections cause synchronisation do not provide such evidence. In strabismic cats, there are no lateral connections between neurons in V1 responding to inputs from different eyes. The reduced synchronisation between these neurons is interpreted as indicating a synchronising role of the lateral connections (König et al, 1993). However, strabismic cats only analyse the visual information of one eye at a time. So, the data can also be interpreted as indicating that (lack of) synchronisation results from the (lack of) processing of the information by higher visual layers. Another set of data shows that severing the callosal link between the visual areas in the two hemispheres eliminates the synchronisation between neurons in these areas (Engel et al., 1991b). This is also taken as evidence for the role of lateral connections for synchronisation. However, callosal fibres also comprise the feedback fibres from higher areas in the contralateral hemisphere. Thus the data could also indicate that synchronisation is mediated by feedback projections from the area that uses the information. This is consistent with findings by (Sillito et al, 1994) and (Munk et al., 1995).