54. REWARD LEARNING IN LATE-LIFE DEPRESSION: A NOVEL LATENT-CAUSE INFERENCE APPROACH

Introduction

Older adults with depression experience deficits in processing socially rewarding experiences. These deficits include blunted reward anticipation – reduced expectation that pleasurable social experiences will occur in the future. Reduced reward anticipation is associated with greater anhedonia severity and lower tendency to seek rewarding experiences. We leveraged latent-cause inference, a computational and conceptual framework for cognitive learning processes, to investigate social reward deficits in late-life depression. We hypothesized that blunted reward anticipation might result from clustering rewarding and non-rewarding events together, instead of accurately distinguishing them. This clustering in turn could lead to anticipation of negative or neutral, rather than positive, social experiences in the future. We aimed to investigate whether depressed older adults, compared to healthy controls, show poorer social reward learning (i.e. reduced segregation between rewarding and non-rewarding social stimuli).

Methods

Fifty-eight participants (mean age: 64.2 years [SD: 7.0], 42 female [72.4%]) were included in the study: 25 older adults with late-life depression (mean age: 65.0 [SD: 7.0] years, 23 female [92.0%]) and 33 healthy controls (mean age: 63.6 [SD: 7.0] years, 19 female [57.6%]). Participants completed our novel “Social Task for Assessment of Reward” (STAR) task four times over 9 weeks (baseline, week 3, 6, and 9). The task consisted of 70 trials: 35 social reward trials (a cue of anticipating social reward feedback); 35 non-reward trials (a cue of anticipating no social reward feedback). We applied mixed-level linear models to investigate differential effects of cue response over time in depressed older adults vs. healthy controls.

Results

We found that both depressed and healthy control individuals showed faster reaction times for anticipated social reward vs. non-reward trials (F1, 13288 = 7.89, p = 0.004989). Further, the depressed group, vs. healthy controls, showed smaller differences in reaction times between social reward and non-reward trials (F1, 13288 = 6.13, p = 0.01333). This effect did not change over time (F1, 13288 = 0.665, p = 0.5737), suggesting persistent diminished segregation of social reward and non-reward trials.

Conclusions

Our findings suggest that depressed older adults, compared to healthy controls, show reduced segregation between rewarding and non-rewarding social stimuli. This lower segregation might indicate poorer social reward learning in late-life depression, specifically reduced anticipation of socially rewarding outcomes. Our results can inform the development of interventions to restore adaptive segregation between rewarding and non-rewarding events to improve social reward learning. Specifically, future work will investigate whether psychotherapy for increasing engagement in socially rewarding experiences improves segregation between rewarding and non-rewarding experience. It could also guide personalization of psychotherapy interventions tailored to individuals’ unique reward-system-supported behaviors.