Rhythms of Association: Pre-stimulus Brain States and the Formation of Multisensory Memory

Abstract

The ability to form memories of multisensory experiences is strongly tied to different types of electrophysiological activity exhibited by different areas of the brain. Among those, theta (3-7 Hz) and alpha (8-12 Hz) oscillations have been shown to be heavily involved in processes of binding information for subsequent memory encoding, as well as inhibiting interfering information in favor of process optimization. However, the brain is always active, exhibiting ongoing oscillatory activity even in the absence of stimulation. The goal of this dissertation was to investigate how these ongoing, pre-stimulus brain states affect the formation of multisensory associations in service to memory. Three empirical studies were conducted in an attempt to shed light on this question. In the first study, participants were required to memorize individual associations between semantically unrelated images and sounds in the context of a Subsequent Memory Effects paradigm. The findings show that later remembered associations exhibited significant increases in theta as well as alpha activity before the stimulus was even presented, suggesting that pre-stimulus low-frequency oscillations are an important correlate of memory success. Based on these results, a modified version of the SME paradigm was used in the second study, in which visual sensory stimulation was utilized to actively modulate theta and alpha activity before encoding happened. The study demonstrated that visual stimulation in the alpha band can significantly increase memory performance as compared to controls, suggesting that pre-stimulus alpha activity may be functionally relevant to the process of encoding multisensory associations. In the third study, the same audiovisual stimulus pairs were presented sequentially rather than simultaneously to assess whether modality order affects subsequent encoding, and whether it can be treated as a contextual feature in the memory trace. The findings suggest a modality-dependent correlation of low-frequency oscillation before and during encoding with subsequent memory performance, suggesting that the order in which sensory information arrives in the brain can affect ongoing and stimulus-evoked brain states. Taken together, these observations point towards a division of labor between theta and alpha oscillations, expanding the notion of binding and inhibition to the domain of ongoing pre-stimulus oscillatory states. The evidence supports the idea that the ability to memorize experiences does not start with having something to experience, but with the state the brain is in before the experience and its encoding begins. Although not without limitations, these insights contribute to our understanding of how the phenomenon of memory comes to be, and may further contribute to future work in clinical and non-clinical settings.