COVID-19 social distancing fundamentally shifted human social communication, encouraging more online interaction alongside increased social isolation and environmental uncertainty. While laboratory studies suggest that these changes might alter reward processing, little is known about how social media reward processing responds to large-scale environmental change. Here, we used COVID-19 social distancing to investigate such adaptation on social media. By fitting computational models to US Twitter data (535 users; 553,945 tweets), we found that social media posting is driven by both reinforcement learning (receiving more ‘likes’ decreased time between posts) and habitual processes (time between posts was influenced by previous behaviour). During social distancing, posting was more driven by reward, with ‘likes’ more strongly shaping posting frequency (β = −0.052, p < .001). COVID-19 social distancing therefore shifted behaviour toward a greater reliance on social media rewards, suggesting that changes in reward learning might underlie behavioural adaptation to environmental disruption.
