Attention to the relations between visual features modulates hippocampal representations. Moreover, hippocampal damage impairs discrimination of spatial relations. We explore a mechanism by which this might occur: modulation by the acetylcholine system. Acetylcholine enhances afferent input to the hippocampus and suppresses recurrent connections within it. This biases hippocampal processing toward environmental input, and should improve externally-oriented, hippocampally mediated attention and perception. We examined cholinergic modulation on an attention task that recruits the hippocampus. On each trial, participants viewed two images (rooms with paintings). On “similar room” trials, they judged whether the rooms had the same spatial layout from a different perspective. On “similar art” trials, they judged whether the paintings could have been painted by the same artist. On “identical” trials, participants simply had to detect identical paintings or rooms. We predicted that cholinergic modulation would improve performance on the similar room task, given past findings that hippocampal representations predicted, and hippocampal damage impaired, behavior on this task. To test this, nicotine cigarette smokers took part in two sessions: one before which they abstained from nicotine for 12 hours, and one before which they ingested nicotine in the past hour. Individual differences in expired breath carbon monoxide levels — a measure of how recently or how much someone smoked — predicted performance improvements on the similar room task. This finding provides novel support for computational models that propose that acetylcholine enhances externally oriented attentional states in the hippocampus.

All .rda R binary data files uploaded here were produced from our final run of the analysis code posted in this GitHub repository. The raw data (uploaded to the GitHub repository, not uploaded here) come from 50 participants completing two sessions of our attention task, one after not smoking cigarettes for the previous 12 hours, and one after smoking cigarettes within the last hour. These intermediate R data files include model outputs from Bayesian multilevel regression analyses using the rstanarm package, and bootstrap-resampled datasets for nonparametric analyses.

Please see the GitHub repository for the raw data, the data dictionary describing the raw data, and the R analysis code that produces these intermediate files.

Note that the code was run without setting a random seed, so subsequent runs of the GitHub code on your local machine to reproduce our analyses from scratch should return values near, but not identical to, those in the files here.