Stimulus duration (sec)

Human adults

FIGURE 7.3 Temporal gradients obtained from the 3-, 5-, and 8-year-olds in a temporal generalization task in two standard duration conditions (4 msec and 4 sec) (Droit-Volet, 2002), with an example of temporal gradient obtained from the human adults derived from Wearden (1992). (From Droit-Volet, S., Q. J. Exp. Psychol., A, 55A, 1193-1209.)

gradients were flatter in the 3-year-olds than in the 5-year-olds and in the 5-year-olds than in the 8-year-olds. At this last age, the gradients were nearly identical to those found in the adults. Furthermore, there was an age-related change in the shape of the temporal generalization gradient. As shown in Figure 7.3, the human adults usually exhibit a right-skewed gradient, so that stimuli longer that the standard are more likely to be confused with the standard than are stimuli shorter by the same amount. However, in our studies, the adult-like gradient has been obtained only at the age of 8 years old; the youngest children, aged 3 and 5, produced a more or less symmetrical gradient, typical to those found in animals. Thus, there is an age-related shift from animal-like symmetrical gradients to adult-like rightward-skewed gradients in children (Droit-Volet, 2002; Droit-Volet et al., 2001). However, McCormack et al., (1999) did not find this developmental shift in the shape of their temporal generalization gradients. Gradients for their 5- and 8-year-olds were significantly skewed to the left, symmetrical gradients being observed at the age of 10 years, and rightward skewed gradients only at adulthood.

In the framework of a systematic analysis of developmental changes in temporal generalization performance, the question is always how to explain these differences between adults and children. Figure 7.4 shows adult-like gradients (black curves) and several theoretical cases of departure (open curves) from these gradients produced by different sources of variation. Of course, these different sources of variation can cumulate, resulting in a particular pattern of timing behavior. Figure 7.4a shows gradients that have the characteristics of those found in adults, but that differ in relative steepness. One gradient is significantly flatter than the other. Flatter gradients indicate lower temporal sensitivity. The first potential source of developmental change is in the temporal sensitivity manifested in the steepness of the gradient. Figure 7.4b also shows adult-like gradients, but one gradient shows an asymmetrical increase in the probability of yes responses for comparison stimulus durations. This is the second developmental possibility: identical shape at different ages, but different threshold levels for responding. Figure 7.4c shows a similar possibility, but now a higher proportion of yes responses occurs also at the shortest comparison stimulus durations. A high proportion of yes responses for the shortest durations is virtually unknown in human adults, who produce a proportion of yes responses near zero for these durations. However, Church and Gibbon (1982) have observed this pattern of responses in rats and have interpreted it as the result of responses emitted at random, and not controlled by signal duration. Therefore, the third possibility is a higher tendency in young children to produce random responses. Figure 7.4d shows the last developmental possibility, with an adult-like rightward asymmetrical gradient and a gradient that is shifted to the left. In this case, stimuli shorter than the standard are confused more with the standard than stimuli longer than it. So, the last developmental possibility is a change in the shape of the generalization gradient.

Our developmental model based on scalar timing theory — a developmental version of the modified Church and Gibbon model (Wearden, 1992) — allows us to take into account these different potential developmental changes in temporal generalization performance. As described above, this model supposes that the amount of pulses accumulated for the just-presented duration is stored in short-term memory. As regards the standard duration, it is stored in long-term memory as a

Coefficient of variation of the long-term memory representation of the standard

Coefficient of variation of the long-term memory representation of the standard

0 0

Post a comment