Functional magnetic resonance imaging of human olfactory working memory

This work discusses olfactory anatomy of the human being and investigational methods used in investigating human olfaction as well as experiments designed to elucidate the functional anatomy of human olfactory working memory (OWM). The impetus for this work arises from observations that olfaction and particularly OWM may be impaired in neurodegenerative diseases. The first experiment was conducted to determine psychometric properties of multiple OWM tasks for future research. Tasks investigated included a paired discrimination task, n-back tasks, and a modified version of the Sternberg task. As task difficulty and working memory load increased, accuracy decreased, and response time increased. Performances on the tasks were highly correlated, suggesting a common cognitive process between tasks. These tasks did not correlate with odor identification or threshold measures, used as screens for general olfactory function. The second experiment investigated the underlying neural substrates of OWM with fMRI. Tasks performed in the scanner included the n-back tasks and the modified Sternberg task from the psychophysical study as well as odor detection tasks. Of the tasks studied, the n-back tasks produced large patterns of activation while the modified Sternberg task provided only minimal activation over odor detection. Contrasts of the olfactory working memory tasks over odor detection tasks produced activation in the dorsolateral prefrontal cortex, dysgranular ventral anterior insula, and posterior lateral parietal cortex. The role of each of these structures is not entirely clear; however, a model is proposed in which the executive processes are mediated by the dorsolateral prefrontal cortex, the left posterior lateral parietal cortex is used to manipulate stimuli, the left dysgranular ventral anterior insula scans and retrieves items from the short-term store for on-line comparison of features, and the right superior ventral anterior insula and posterior lateral parietal function as a chemosensory working memory slave system. Hippocampal voxels existed for both n-back tasks in which activity correlated with task accuracy though there were no significant hippocampal activations from the cognitive subtractions. Thus, the role of the hippocampus in OWM, if any, remains unclear.