Single units and sensation: a retrospect

The single-neuron doctrine is reexamined, and the search for causal links between single units and sensation reviewed. Although several decades of single-unit recording have been very successful in elucidating physiological mechanisms, linking signals from a single cell and perception has progressed at a slower rate. Nevertheless, analysing the activity of single neurons has achieved significant gains and remains the most promising level for elucidation of processing algorithms in the visual system. At the subcortical level, the conclusion that signals from just a single cell can carry enough information for some kinds of performance remains (almost) valid. Under carefully controlled conditions, just a few impulses in a few retinal ganglion cells are an adequate signal for the cortex to initiate a behavioural response. Elucidating cortical codes has been more difficult, and evidence exists suggesting the sharing of responsibility for a task among cell assemblies; how large these assemblies are, and how to test for them neurophysiologically, remains a challenge.