Brain mechanisms involved in selective attention in humans can be studied by measures of regional blood flow and metabolism (by positron emission tomography) which help identify the various locations with enhanced activities over a period of time of seconds. The physiological measures provided by scalp–recorded brain electrical potentials have a better resolution (milliseconds) and can reveal the actual sequences of distinct neural events and their precise timing. We studied selective attention to sensory inputs from fingers because the brain somatic representations are deployed over the brain convexity under the scalp thereby making it possible to assess distinct stages of cortical processing and representation through their characteristic scalp topographies. In the electrical response to a finger input attended by the subject, the well–known P
300
manifests a widespread inhibitory mechanism which is released after a target stimulus has been identified. P
300
is preceded by distinct cognitive electrogeneses such as P
40
, P
100
and N
140
which can be differentiated from the control (obligatory) profile by superimposition or electronic subtraction. The first cortical response N
20
is stable across conditions, suggesting that the first afferent thalamocortical volley is not affected by selective attention. At the next stage of modality–specific cortex in which the sensory features are processed and represented, responses were enhanced (cognitive P
40
) only a very few milliseconds after arrival of the afferent volley at the cortex, thus documenting a remarkable precocity of attention gain control in the somatic modality. The physiology of selective attention also provides useful cues in relation to non–target inputs which the subject must differentiate in order to perform the task. When having to tell fingers apart, the brain strategy for non–target fingers is not to inhibit or filter them out, but rather to submit their input to several processing operations that are actually enhanced when the discrimination from targets becomes more difficult. While resolving a number of such issues, averaged data cannot disclose the flexibility of brain mechanisms nor the detailed features of cognitive electrogeneses because response variations along time have been ironed out by the bulk treatment. We attempted to address the remarkable versatility of humans in dealing with their sensory environment under ecological conditions by studying single non–averaged responses. We identified distinct cognitive P
40
, P
100
, N
140
and P
300
electrogeneses in spite of the noise by numerically assessing their characteristic scalp topography signatures. Single–trial data suggest reconsiderations of current psychophysiological issues. The study of non–averaged responses can clarify issues raised by averaging studies as illustrated by our recent study of cognitive brain potentials for finger stimuli which remain outside the subject'sawareness. This has to do with the physiological basis of the ‘cognitive unconscious’, that is, current mental processes lying on the fringe or outside of phenomenal awareness and voluntary control, but which can influence ongoing behaviour. Averaged data suggest that, in selective auditory attention, the subject may not notice mild concomitant finger inputs. The study of non–averaged responses documents the optional and independent occurrence of the cognitive P
40
, P
100
and N
140
(but not P
300
) electrogeneses while the finger inputs remain outside phenomenal awareness. These results suggest that the subject unconsciously assigns limited cognitive resources to distinct somatic cortical areas thereby submitting finger inputs to an intermittent curtailed surveillance which can remain on the fringe or outside consciousness. The study of cognitive electrogeneses in single non–averaged responses is making possible a neurophysiology of cognition in real time.
Phenomenal awareness can emerge without attention