Tracking selective attention in a musical cocktail

Abstract

The cocktail party effect refers to mankind’s ability to focus attention on a single sound within noisy or complex auditory environments, in order to extract the most behaviourally-relevant information present. To investigate this cognitive phenomenon in a precise manner, we used frequency-tagging to separate neural auditory steady-state responses (ASSRs), which are specific to the driving frequency of each auditory source, from a neural mix elicited by multiple simultaneous stimuli. Although this technique has been utilized before in previous research, results were inconsistent on how the ASSR is influenced by selective attention, with findings based primarily on attention being directed to either ear rather than to sound features. In the current doctoral project, a mixture of melody streams was presented to both ears identically (diotically) as participants directed attention selectively to only one stream while following the development of its pitch contour. Our main aim was to examine if selective attention to the pitch and timing of frequency-tagged musical tones influences the corresponding ASSR. In addition, bottom-up attention towards musical tones was also manipulated with salient changes in pitch. Using magnetoencephalography (MEG), we showed that the attentional enhancement of the ASSR can already be observed at general sensor level.

Another goal of this project is to characterize the ASSR in terms of its source distribution, properties, and attentional modulation, during and after auditory stimulation. We also assessed how some of these factors change according to the complexity of the stimulus cocktail by comparing the ASSR when the musical streams simultaneously overlapped with one another and when they were completely separated in time. Distributed source analyses revealed multiple ASSR sources distributed across the frontal, parietal and temporal cortices, with each of these areas modulated to different extents by selective attention. Notably, we uncovered the existence of ASSR attentional modulation outside the temporal lobe, in the bihemispheric frontal, parietal and insular lobes, which has not been previously reported in literature. Our results indicated that the ASSR enhancement from top-down driven attention was strongest at the frontal lobe compared to the temporal and parietal regions, whereas that of bottom-up driven attention was dominant at the right temporal lobe. The ASSR also displayed sensitivity towards individual musicality, demonstrating positive correlations between musical sophistication and ASSR power, as well as between musical sophistication and the degree of ASSR attentional modulation at the parietal cortices, an area commonly associated with musical training.

Surprisingly, we discovered stark differences between the ASSR during and after the stimulation period. While the ASSR during stimulation was affected by the stimulus properties (i.e. volume and pitch), selective attention, and participants’ level of musical sophistication, the post-stimulus ASSR was not influenced by any of these factors. Furthermore, while the ASSR during stimulation was generated primarily from temporal sources, the post-stimulus ASSR originated mainly from the frontal cortex. These findings challenge the notion that the post-stimulus ASSR is merely a continuation of the ASSR during stimulation. Rather, they suggest that the post-stimulus ASSR is an internally-driven signal that develops from an initial sensory processing state generating the ASSR during stimulation, with both types of ASSRs clearly differing in cortical representation and character.

With two simultaneously overlapping streams, we observed suppression of the ASSR power which varied across cortical lobes and hemispheres. This gave rise to a shift in the ASSR source distribution from temporal-dominance during single-stream perception, to proportionally stronger activity in the frontal and centro-parietal cortical regions when listening to simultaneous streams. With the accumulation of evidence highlighting the differences between ASSR sources in the frontal and temporal regions, our results collectively advocate that these sources are characteristically unique, functionally distinct and largely independent from one another.

Taken together, our findings revealed new aspects of the ASSR and ways to effectively extract the modulation effect by selective attention. The ability to readily capture attentional changes in a stimuli-precise manner makes the ASSR a useful tool for investigating selective auditory attention and its interaction with individual long-term auditory experience such as musical training. Importantly, this work advocates a novel ‘beyond the temporal cortex’ perspective on ASSR modulation and is a step forward to studying human cognition in more complex and naturalistic soundscapes using frequency-tagging.