A regular relationship between old and young adults' latencies on their best, average and worst trials

The relationship between corresponding latencies of old and young adults early in practice are examined across a range of tasks (choice reaction time, memory scanning and classification) in a within‐subjects design. These relations are well described by linear functions in the case of cognitive processing times (CTs) on all tasks, movement times (MTs) from choice reaction and viewing times (VTs) from memory scanning. This linearity holds not only for comparisons of medians but also for other points of the within‐subject distribution (WSD), specifically the 10th, 25th, 50th, 75th and 90th percentiles. The data for MTs lie above the CT points, suggesting relatively greater slowing of movement than of cognition in the aged. Slopes for VTs decrease with set size, indicating less compensation for increased memory load by (he aged. Most importantly, one linear equation describes 97% of the relationship between old and young people's CTs across all tasks and percentiles. This strongly suggests that the relations between CTs for all percentiles and tasks are collinear, at least early in practice. This result could indicate that trial‐by‐trial fluctuations in individual performance are quantitatively similar to changes in performance due to task difficulty with both being affected by age in a similar fashion. Further, it provides strong evidence for a unitary effect of age on cognitive processing speed and against models which invoke extra components or occasional unusual trials by the aged to explain their slower typical performance. 1988 Australian Psychological Society