Assessing behavioral control across the adult lifespan using a novel outcome revaluation task

Abstract

While many studies have investigated the effects of aging on cognition, relatively few have examined aging impacts on habitual behavioral control. To assess these relationships, 151 adults across the lifespan (47.09 ± 17.17 years old, range = 19-80) completed a novel instrumental outcome revaluation task, where participants made keyboard responses to abstract stimuli to gain digital currency before completing a revaluation test where the outcome of one stimulus was negatively altered while the other retained its value. Participants also completed questionnaires relating to psychiatric symptoms. Habitual responding was measured in terms of the response rate to the revalued stimulus relative to the response rate to the stimulus that was not revalued. There were significant positive effects of obsessive-compulsive symptoms and significant negative effects of depressive symptoms on habitual behavior. In addition, results revealed a modest effect of chronological age on habitual behavior These results indicate subtle changes in behavioral control across the adult lifespan and support previous work showing that certain psychological measures including obsessive-compulsive symptoms are associated with increased habitual responding.

Keywords: Aging; Behavioral control; Depression; Habits; Obsessive-compulsive behavior.

Fig. 1

Schematic of the appetitive outcome revaluation task. Participants were tasked with viewing abstract stimuli and making an associated response (II or PP) to gain currency (+$5.00) and beat a virtual rival (300 trials, 100 per abstract stimulus; participant counter – top, rival counter – bottom). Following an initial instrumental learning phase, one of these stimuli was revalued, now associated with monetary loss (-US $5.00) instead of gain (+ US $5.00). Participants had to figure out which stimulus was revalued on their own (12 trials, 4 per stimulus). Participants then completed a test phase under these revalued conditions, where they were told to keep responding to the unaltered stimulus but not the revalued stimulus, while their currency counters were broken (30 trials, 10 per abstract stimulus; 342 trials total).

Fig. 2

Partial regression plot depicting age effects on correct responding at test. Despite a significant age x stimulus valuation interaction, simple slopes were not significant, though the effect of age on revalued stimulus responding approached significance (top). Shaded regions = 95% CIs. The ‘| Others’ terminology clarifies that these graphs partial out the variance of the covariates in the referenced model, thus residuals are shown.

Fig. 3

Partial regression plots depicting obsessive-compulsive and depressive effects on correct responding at test. Obsessive compulsive symptoms had a significant positive effect (left) while depressive symptoms had a significant negative effect on correct responses at test (right). Shaded regions = 95% CIs. The ‘| Others’ terminology clarifies that these graphs partial out the variance of the covariates in the referenced model, thus residuals are shown.

Fig. 4

Adjusted probability plots of habitual status (non-habitual, habitual) by age and obsessive-compulsive symptoms. While age did not significantly predict habitual status (left), obsessive-compulsive symptoms was a significant positive predictor (right), such that for each unit increase in obsessive-compulsive symptoms, the odds of responding habitually increase by 4.4%. Shaded regions = Wald 95% CIs. Tick marks represent observations for positive (top, y = 1) and negative (bottom, y = 0) residuals.

Fig. 5

Partial regression plots depicting effects of age and habitual status on various reaction time measures. Habitual status, but not age, significantly predicted average learning and test phase reaction times (sec.), with habitual participants responding significantly slower than non-habitual participants. Age, but not habitual status, significantly predicted the learning and test phase subtraction measures, indicating that older adults had larger gaps between the first and second responses than younger adults. Shaded regions = 95% CIs. The ‘| Others’ terminology clarifies that these graphs partial out the variance of the covariates in the referenced model, thus residuals are shown.