TEMPORAL OSCILLATIONS lN HUMAN PERCEPTION

The notion that human perceptual decisions are based on discrete processing cycles rather than a continuous accumulation ofinformation was examined experimentally. Significant periodicities were found in hwnan response times (RT) to feature and conjunction discrimination tasks in the visual and auditory modalities. Individual RT histograms were multimodal. with regularly spaced peaks and troughs. indicating that responses were emitted more frequently at regularly recurring time intervals following stimulus presentation. On average, responses were initiated after four to seven discrete processing steps whose "quantum" duration Il'as proportional to task difficulty. Measurement of human response times (RT) is a widespread technique in cognitive psychology (Luce, 1986; Posner, 1978). It is commonly assumed that mental processing time has a large stochastic component and may vary continuously, on a millisecond scale, as a function of stimulus characteristics and task demands (Luce, 1986). Psychological models postulate a series of consecutive processing stages (Donders, 1969; Sternberg, 1969), each with variable duration. RT, which is the sum of these durations, is therefore expected to be broadly distributed, and in general only its mean or median is considered significant. A number of authors have considered the alternative possibility of a temporal quantification in mental processing. Stroud (1955) proposed the "perceptual moment hypothesis," according to which perceptual inputs are chunked into discrete temporal intervals according to the regular oscillations of a "central intermittence." Since then, data suggestive of an "internai dock" or subjecThe author is on leave at the Institute of Cognitive and Decision Sciences. Straub Hall, University of Oregon, Eugene, OR 97403, where correspondence should be addressed; e-mail: [email protected]. 264 tive "time quanta" have been reported occasionally (Allport, 1968; Augenstine, 1955; Collyer, Broadbent, & Church, 1992; Rarter & White, 1968; 10keit, 1990; Kristofferson, 1967a, 1967b, 1980; Latour, 1967; Michon, 1967; Pôppel, 1970; Shallice, 1964; M. Treisman, Faulkner, Naish, & Brogan, 1990; M. Treisman, Faulkner, & Naish, 1992). Several of the se studies have reported -multiple discrete peaks in histograms of a few hundred reaction times or ocular movement times. The observed periodicities have sometimes been tentatively linked to physiological events, such as the alpha rhythm of the electroencephalogram (Callaway & Yeager, 1960; Kristofferson, 1967a, 1967b; Latour, 1967). Most of those data were collected before computer facilities existed. and their statistical reliability has been criticized severely (Vorberg & Schwarz, 1987; Vroon, 1970, 1974). However, more recently, eIectrophysiological recordings from auditory and visual brain areas have revealed stimulus-dependent oscillations in the range of 30 to 80 Hz (BasarEroglu & Basar, 1991; Eckhorn et al., 1988; Engel. Kônig, Kreiter, & Singer, 1991; Engel, Kreiter, Kônig, & Singer, 1991; Engel, Kônig, & Singer, 1991; Galambos, Makeig, & Talmachoff. 1981; Gray, Kônig, Engel, & Singer, 1989; Gray & Singer, 1989; Pantev et al., 1991; Ribary et al., 1991). These results have again suggested that the neuronal encoding of sensory information may be discrete in time. The present study investigated whether a temporal periodicity or discreteness might be perceptible in individual subjects' distributions of RTs during perceptual discrimination tasks. To this end, 1,600 RTs were collected from each of five highly trained observers in each of four two-choice tasks. It was reasoned that the period or phase of the putative oscillatory processes might differ between subjects or between tasks, thereby hindering any meaningful analysis of composite data sets. The experiCopyright (