The effects of aging on the interaction between reinforcement learning and attention

It is unknown whether healthy aging changes how people combine learning with selective attention when a task is cognitively demanding. We compared younger and older adults in two experiments where only one visual dimension predicted reward and participants had to discover this dimension by trial and error. In some cases, we gave hints about what to focus on. Both age groups behaved like reinforcement learners who use attention to narrow the world, but modeling suggested older adults applied a tighter filter (updating values about fewer features at once). That difference in cognitive strategies account for when older adults were less accurate at obtaining reward.