Abstract
Flexible instrumental learning is required to harness the appropriate behaviors to obtain rewards and to avoid punishments. The precise contribution of dopaminergic midbrain regions (substantia nigra/ventral tegmental area [SN/VTA]) to this form of behavioral adaptation remains unclear. Normal aging is associated with a variable loss of dopamine neurons in the SN/VTA. We therefore tested the relationship between flexible instrumental learning and midbrain structural integrity. We compared task performance on a probabilistic monetary go/no-go task, involving trial and error learning of: “go to win,” “no-go to win,” “go to avoid losing,” and “no-go to avoid losing” in 42 healthy older adults to previous behavioral data from 47 younger adults. Quantitative structural magnetization transfer images were obtained to index regional structural integrity. On average, both some younger and some older participants demonstrated a behavioral asymmetry whereby they were better at learning to act for reward (“go to win” > “no-go to win”), but better at learning not to act to avoid punishment (“no-go to avoid losing” > “go to avoid losing”). Older, but not younger, participants with greater structural integrity of the SN/VTA and the adjacent subthalamic nucleus could overcome this asymmetry. We show that interindividual variability among healthy older adults of the structural integrity within the SN/VTA and subthalamic nucleus relates to effective acquisition of competing instrumental responses.
Highlights
To efficiently harvest reward and avoid punishment, humans need to learn appropriate instrumental responses (Dickinson and Balleine, 2002) (O’Doherty et al, 2004)
We found a main effect of action indicating that participants were better at learning go compared to no-go choices (F1,41 1⁄4 7.29, p 1⁄4 0.01)
Our results reveal that some healthy older adults are unable to flexibly learn 2 responses for reward within a single task
Summary
To efficiently harvest reward and avoid punishment, humans need to learn appropriate instrumental responses (Dickinson and Balleine, 2002) (O’Doherty et al, 2004). Healthy young human adults readily learn to act to obtain a reward, or not to act to avoid a punishment, they have difficulties learning to act to avoid a punishment and not to act to obtain a reward (Guitart-Masip et al, 2012b) This inflexibility in learning suggests that signals that predict rewards are prepotently associated with behavioral activation promoting approach behavior, whereas signals associated with punishments are intrinsically coupled to behavioral inhibition promoting avoidance. These behavioral tendencies can be described as Pavlovian biases that corrupt the flexibility of. Computational modeling in young adults has shown that the observed pattern of behavior is captured by a model incorporating a Pavlovian bias, where the strength of this bias is related to failure to learn the conflicting conditions: no-go to win and go to avoid losing (Guitart-Masip et al, 2012b)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
