In previous studies, it has been demonstrated that visible persistence-the period for which the perceived duration of a stimulus exceeds its physical duration-can be extended by briefly incrementing the luminance of the stimulus immediately prior to offset. Using a two-component pattern integration task, we show that this effect is not an artifact of change in the total luminous flux within the stimulus. Visible persistence was unaffected by overall luminance of the stimulus. It was also time-locked to the luminance increment. Visible persistence is seen to result from a process that is initiated by stimulus onset and that can be either wholly or partially reinitiated by the onset ofthe luminance increment. The duration of this process (which determines the duration of stimulus visibility) can be modified in a graded fashion by stimulus events that occur after its initiation. We outline a single-process inhibitory feedback model of the persistence mechanism that accounts for the present findings.Visual displays presented for less than about 100 msec remain visible for some interval after the display has physically terminated (Efron, 1970a(Efron, , 1970b. The additional period of visibility is known as visible persistence (Coltheart, 1980). Visible persistence is characterized by stimulus dependencies that differentiate it from other well-studied visual aftereffects such as retinal afterimages (Coltheart, 1980; Di Lollo, Clark, & Hogben, 1988), and from well-known forms of short-term memory such as iconic memory (Coltheart, 1980). In particular, iconic memory and afterimages are known to increase in duration as the energy of the inducing stimulus is increased, by increments in either its luminance or its duration. By contrast, it is well established that the duration of visible persistence is inversely related to the duration of the stimulus (the inverse duration effect of Coltheart, 1980) and that visible persistence is usually inversely related to stimulus luminance (the inverse intensity effect).Various general accounts of the mechanism of visible persistence have been proposed. Di Lollo and Bourassa (1983) outlined two that they regarded as equally tenable. The first of these-the processing hypothesis-derives from the sensory-coding hypothesis put forward by Di Lollo (1980). In essence, this holds that persistence is a product of the activity of sensory-coding mechanisms that operate at the earliest stages of visual processing to recode the retinal image in terms of primitive features such as edges, bars, dots, and discontinuities. According to the processing hypothesis, the duration of the activity that a stimulus initiates in the persistence mechanism corresponds to the duration of very early stages in a chain of processing events triggered by a change in visual stimulation. Stimuli persist to the extent that this activity outlasts the corresponding physical stimulation.The second of the two accounts outlined by Di Lollo and Bourassa (1983) was proposed initially by Galifret (1977a, 1977b) and subsequen...