Are There Age-Related Differences in the Ability to Learn Configural Responses?

Abstract

Age is often associated with a decline in cognitive abilities that are important for maintaining functional independence, such as learning new skills. Many forms of motor learning appear to be relatively well preserved with age, while learning tasks that involve associative binding tend to be negatively affected. The current study aimed to determine whether age differences exist on a configural response learning task, which includes aspects of motor learning and associative binding. Young (M = 24 years) and older adults (M = 66.5 years) completed a modified version of a configural learning task. Given the requirement of associative binding in the configural relationships between responses, we predicted older adults would show significantly less learning than young adults. Older adults demonstrated lower performance (slower reaction time and lower accuracy). However, contrary to our prediction, older adults showed similar rates of learning as indexed by a configural learning score compared to young adults. These results suggest that the ability to acquire knowledge incidentally about configural response relationships is largely unaffected by cognitive aging. The configural response learning task provides insight into the task demands that constrain learning abilities in older adults.

Fig 1. Initial stimulus response mappings phase: Example trial from the stimulus-response mappings phase that preceded learning trials.

Letters under each face stimulus correspond to the letters on the keyboard mapped to each face stimulus (left hand: q, w, e, r; right hand: u, i, o, p).

Fig 2. (A) Example trial sequence for one session of configural learning trials. Letters under each face stimulus on the first screen correspond to the letters on the keyboard mapped to each face stimulus (left hand: q, w, e, r; right hand: u, i, o, p); (B and C) Matrix of the two possible response mappings (i.e., for counterbalancing purposes), where the number in each cell refers to the frequency at which the specific face-face pair appeared in the task for a given subject (280 = Frequent trial pairing and 40 = Infrequent probe trial pairings). The numbers in the cells demonstrate that in both response mapping sets (B and C), a given face is seen 640 times (i.e., sum of any row or column) and thus all individual response elements are equally familiar across the learning task.

Fig 3. Proportion correct for frequently performed (solid lines) and infrequently performed (dotted lines) pairs across session for (A) day 1 and (B) day 2.

Shaded regions represent standard error. Note in linear model, the time variable (session) was centered so that the linear slope for each group was calculated at the midpoint of each day (session 3).

Fig 4. Reaction times for frequently performed pairs on (A) day 1 and (B) day 2.

Only frequently performed pairs are modeled, rather than the comparison of frequent and infrequent pairs; this served as a method for assessing motor skill learning (basic response speeding) separately from configural learning. Participants performed 7 blocks of frequent pair responses in each of 5 sessions per day, which are plotted sequentially in the figure above as blocks 1 through 35 for Day 1 and Day 2. Dotted lines represent block-wise group averages; shaded region represents standard error; solid lines represent model-predicted values.

Fig 5. Observed and modeled configural learning scores plotted across session for (A) day 1 and (B) day 2.

Dashed lines connect group average learning scores for each session; shaded regions represent standard error; solid lines represent model-predicted values. Note in linear model, the time variable (session) was centered so that the linear slope for each group was calculated at the midpoint of each day (session 3).