produced interval

FIGURE 9.8 Pulse accumulation in conditions of low and high frequency of trials with breaks. Accumulation is faster in the low- than in the high-frequency condition, because attentional shifts are less frequent. The criterion is reached earlier when accumulation is faster, resulting in relatively short produced intervals.

triggers the end of interval production. The difference in accumulated counts at the end of prebreak duration (a in Figure 9.8 when break location is 2200 msec) would determine the difference between produced intervals (a') in the high- and low-frequency conditions. Because the temporal criterion is reached earlier in the low-than in the high-frequency condition, produced intervals are shorter. The difference would be proportional to prebreak duration: we can deduce from Figure 9.8 that the difference in accumulated counts between the low- and high-frequency conditions would be smaller if the break occurred earlier, for example, 700 msec after the first key press rather than 2200 msec. Differences between produced intervals in the low- and high-frequency conditions are thus proportional to prebreak duration, which results in the observed interaction between prebreak duration and frequency of trials with breaks.

The relation between produced intervals and prebreak duration is linear in both high- and low-frequency conditions, suggesting that the duration and frequency of shifts are relatively stable throughout the prebreak duration. Of course, even though shifts are equally spaced and of the same duration for illustrative purposes in Figure 9.8, we do not assume this to be the case.

In this interpretation, accumulation proceeds according to an ON-OFF operation mode, depending on whether time is selected as the focus of attention. Accumulation would be under control of a mechanism compatible with a flickering switch, discussed in recent articles on the role of selective attention in timing (Lejeune, 1998, 2000; Penney et al., 2000; Zakay, 2000). In the information-processing model of interval timing (Church, 1984; Gibbon and Church, 1984; Gibbon et al., 1984), pulses are accumulated when the switch is closed, and accumulation is interrupted when the switch is opened. Given the constraints of a production task, the switch is normally closed during the interval production, allowing pulse accumulation. The state of the switch is modified in two periods of interval production with breaks. First, it is opened during the break itself, interrupting accumulation (e.g., Meck et al., 1984; Roberts and Church, 1978; Roberts, 1981). Second, it alternates between closed and opened states during the prebreak period (Fortin and Massé, 2000). The results from the present experiment suggest that the level of expectancy for a break controls the frequency of changes in the state of the switch during the prebreak period: the switch would be opened more frequently when a break is strongly expected. Note, however, that longer openings of the switch would produce the same results.

The slope of mean productions as a function of prebreak duration is almost three times greater in the high-frequency condition than in the low-frequency condition (0.21 vs. 0.08). This could be taken as an estimate of the relative frequency (or duration) of OFF periods in the two conditions. On average, OFF periods would be three times more frequent (or longer) in the high- than in the low-frequency condition.

Mean intervals in trials with no breaks, shown in Figure 9.7, seem to follow the linear trend of the production function in the low-frequency condition, but not in the high-frequency condition. This may be related to the difference in proportion of trials with and without breaks. Because almost all trials included a break in the high-frequency condition, participants could develop a precise representation of the four possible break locations so that when the last possible location was past, they stopped expecting a break. That would explain why productions are not longer than they were at the last break location in the high-frequency condition. In contrast, a minority of trials included a break in the low-frequency condition. Representation of the possible break locations was therefore probably much less accurate, making it more likely that participants were still expecting a break after the last possible break location. In these conditions, productions should be longer in the no-break trials than in trials at the longest prebreak duration, which was observed in the low-frequency condition.

Given that the primary interest here was to identify the attentional mechanisms involved in break expectancy, the focus of data analysis in experiment 1 was on the break location effect. Break duration was therefore not included in our analysis.

Overall, the results of experiment 1 replicate and extend those obtained in Fortin and Masse's (2000) study. As in this study, we noted in postexperimental interviews that participants were unaware of the effect of prebreak duration. This suggests that time-sharing during break expectancy is not under voluntary control.

0 0

Post a comment