Response times to different spatial frequencies: is there a 100-msec role?

Abstract

Stanley, Howell, and Smith (1980) present an alternative analysis of the persistence data obtained by Bowling and Lovegrove (1980). They argue that stimulus onset asynchrony (SOA) is a more appropriate measure of persistence than interstimulus interval (lSI). In addition, they propose that the use of this measure alters the conclusion made by Bowling and Lovegrove (1980) that the stimulus duration at which persistence ceases to decline steeply with increasing stimulus duration increases with spatial frequency and is related to temporal integration. We consider that SOA is a measure of the total duration of the response to a stimulus, rather than of persistence, and use of this measure does not affect the major conclusions of Bowling and Lovegrove (1980). Stanley et al. (1980) also mentioned that a nonlinear continuous function may provide a better fit to Bowling and Lovegrove's data than a two-limb linear function. This possibility was evaluated by fitting linear, quadratic, and logarithmic functions to the data by the method of least squares. The sums of the squared deviations of the data points from each of these regression relationships and from the two-limb linear relationship were calculated. Table 1 shows that, for each spatial frequency, the deviation of the data points from the two-component linear relationship was lower than that for any other relationship. The two-limb relationship is therefore a better fit to the data than any of the other possibilities investigated. The question as to whether SOA or lSI is the more appropriate measure will be considered in terms of the temporal property measured by these. Using arrays of small light flashes, Di Lollo (1977, 1980) has provided data indicating that the response to a brief stimulus commences at onset and continues for approximately 100 msec (the l00-msec rule?). That is, the total duration of the response produced by a brief stimuli is approximately constant. In the measurement technique employed by Bowling and Lovegrove (1980), the SOA is consequently a measure of this total response duration (and of the duration of any persistence occurring apart from this). That part of the response that occurs during the actual presence of the stimulus cannot be considere~ to be persistence, however, since, by definition, pers~stence is a continued visible representation of the stimulus occurring after its offset. Persistence is consequently the difference between SOA and stimulus durationthe lSI. Di Lollo's data show that the total response duration remains approximately constant at about 100 msec, Consequently, the persistence durations of stimuli shorter than this decline linearly with increasing stimulus duration with a slope of -1.0, a result also obtained by Efron (1970). This approximately constant response duration may be a further manifestation of the limited temporal resolution of the visual system, of which the best known example is Bloch's law. Under conditions of perfect temporal integration, the response duration of stimuli shorter than the critical duration would be invariant with stimulu duration. There are, however, some cases in which perfect reciprocity does not occur at threshold. These include large field size (Owen, 1972) and patterned stimuli (Breitmeyer & Ganz, 1977; Legge, 1978). Under these conditions, the total response duration may increase slightly with increasing stimulus duration up to the critical duration. In this case, the slope of the initial portion of the relationship between persistence and stimulus duration would be more shallow than -1.0, as Bowling and Lovegrove (1980) observed. Stanley et al. (1980) have argued that use of SOA as the dependent variable produced an approximately constant relationship between SOA and stimulus duration at brief stimulus duration for each spatial frequency used. This is not surprising, in view of the relationship between persistence and stimulus duration discussed above. The SOA obtained with brief stimuli did, however, increase slightly with increasing stimulus duration, reflecting the fact that the initial slope of the persistence by duration relationship was not -1.0. Stanley et al. (1980) argue that their reanalysis alters the conclusion of Bowling and Lovegrove (1980) that the duration at which a change in the slope of this relationship occurs is dependent upon spatial frequency, and consequently, that one component is a manifestation of the same processes underlying temporal integration at threshold. They