How attention and context affect neural representations of sounds
Our environments are rich in auditory stimulation and we’re constantly exposed to sounds that vary greatly. And yet, young normally hearing listeners are able to recognize a friend’s voice in a loud bar as well as in a quiet library; they can separate two talkers who speak at the same time, even though only a mixture of the two acoustic signals reaches their ears; and they can anticipate when the next tone in a musical piece occurs based on the music’s temporal structure. We investigate neural correlates of this adaptive auditory system using electroencephalography and functional imaging. In particular, we’re interested in how the context in which a participant hears a sound affects its neural representation (e.g. depending on the sounds that have been presented before it). We’re also interested in how directing attention to a particular sound affects its representation (e.g. depending on whether you’re paying attention to the person who’s talking on your left or right side).
The advantage of listening to a familiar voice
In challenging acoustic environments, even young listeners with normal hearing have difficulty understanding speech. For example, when there’s background noise that’s louder than the speech of interest. However, in these challenging acoustic environments, listeners are better at understanding speech if one of the talkers in the room is someone who they’re familiar with—for example, a friend or a partner (e.g. Johnsrude et al., 2013).
We aim to elucidate the mechanisms by which familiar voices improve speech intelligibility, particularly in challenging listening conditions, such as when multiple people speak simultaneously. We’re interested in the following questions: How do voices become familiar? Are the brain representations of familiar voices similar to, or different from, those of novel voices? To what extent can experience with particular voices be beneficial for listeners with hearing loss? We aim to address these questions using electroencephalography, functional imaging, and behavioural measures.
When speech is clear and presented in quiet, we’re able to understand it without much effort. However, when speech is degraded it becomes more effortful to comprehend (e.g., Wild et al., 2012). We’re interested in unpacking this 'effortful listening' into the cognitive processes (such as memory or attention) that facilitate the perception of degraded speech. We attempt to isolate these cognitive processes by examining speech perception under the simultaneous load of different cognitive tasks—for example, participants may be asked to listen to speech, while maintaining other (auditory or visual) information in memory. This research has implications for understanding the cognitive load of hearing impairment.
Perceptual and neural consequences of aging
Older adults tend to experience particular difficulty listening in challenging and/or changing acoustic environments—for example, when multiple conversations are happening in the same room.
We seek to understand the brain mechanisms that enable us to flexibly listen in various acoustic contexts and how these mechanisms and perceptual consequences change with age. We investigate neural and behavioral correlates of aging using electroencephalography, functional imaging, and psychophysics.
Parcellating the human auditory cortex
Functional magnetic resonance imaging (fMRI) can be used to detect neural activations generated by different areas of the brain. However, the precise anatomical location of activation tends to vary across individuals.
We aim to relate functional activity to anatomical structure using machine learning techniques, which aim to classify anatomical regions based on brain activation patterns. We currently use data obtained from monkeys (macaques) and, in the future, we will apply our algorithms to human brain activity. This project aims to expand our understanding of how brain activity is modulated by intrinsic neural structure.
Localization of epileptic seizure
When medication fails to control epilepsy, surgical resection of the seizure focus may be considered. However, localization of the seizure focus is particularly challenging when no clear structural lesion can be identified. Individuals who don’t show clear lesions on an MRI scan must undergo more extensive evaluation using tools that are costly and invasive. Furthermore, those individuals are more likely to have poor outcomes following surgery.
We’re currently testing an fMRI paradigm with epilepsy patients, which is designed to detect brain abnormalities that may be otherwise invisible with existing assessment tools. If successful, this paradigm could be used as a fast and non-invasive presurgical assessment tool for epilepsy patients.