Shortinterval Timing

A minimum of three target intervals is required to test the basic description of the linear-timing hypothesis. When more than three intervals have been tested, the absolute spacing between conditions generally increased as the magnitude of intervals increased (e.g., Fetterman and Killeen, 1992). Increasing spacing between conditions, as in a geometric series, is useful for evaluating a generalized Weber function because the close spacing between conditions for the shortest intervals permits the documentation of the predicted curve in the data for these short intervals. However, this approach is less appropriate for evaluating departures from a theoretical function, particularly if the temporal locations of departures from a linear function are not known a priori. In this case, it is necessary to test many closely spaced intervals to detect a local departure from a theoretical function; a systematic, local departure from a theoretical function would not be detected if fewer and more widely spaced interval conditions were examined (e.g., Collyer et al., 1992, 1994; Crystal, 1999, 2001b; Crystal et al., 1997; Kristofferson, 1980, 1984).

Data from temporal discrimination procedures using many closely spaced target intervals are shown in Figure 3.1. The data suggest that multiple temporal ranges are characterized by local maxima in sensitivity to time. The data were obtained from a titration procedure with rat subjects (Crystal, 1999, 2001b). For example, in a discrimination task, a "short" or "long" noise stimulus was presented followed by the insertion of two levers in an operant box. Different responses (left- or right-lever press) were required after short or long stimuli to obtain a food pellet. For a given short-duration condition, the duration of the long signal was adjusted (i.e., titrated) after blocks of discrimination trials to maintain discrimination accuracy at 75% correct, which resulted in a long duration approximately 2 to 2.5 times the short duration. Sensitivity to time was measured using the signal detection theory (Mac-millan and Creelman, 1991). Sensitivity to time was approximately constant across short durations of 2 to 34 sec. However, local maxima in sensitivity were observed at approximately 12 and 24 sec (Figure 3.1A and B). When short durations in the millisecond range were examined, local maxima were observed at 0.3 and 1.2 sec (Figure 3.1C).

Local maxima in sensitivity to time violate the linear-timing hypothesis. This phenomenon demonstrates that certain intervals are timed with greater relative precision than other, nearby intervals. Local maxima in sensitivity to time are consistent with an oscillator representation of time in which the location of a local maximum identifies the period of an oscillator. A multiple-oscillator mechanism and other theories are discussed in the section below on implications for theories of timing.

An oscillator interpretation of short-interval timing is supported by recent observations of periodic behavior in the short-interval range. When food is contingent on lever pressing after a random interval, rats search periodically for food (Broadbent, 1994). Thus, behavior was periodic in the absence of periodic stimuli, similar to a free-running rhythm. Furthermore, conditions that promote periodic behavior in the absence of periodic stimuli have recently been identified in the short-interval range (Broadbent, 1994; Kirkpatrick-Steger et al., 1996; Machado and Cevik, 1998).


0 0

Post a comment