Auditory Working Memory

Auditory working memory (AWM) is the process of keeping representations of auditory objects in mind for short duration when the sounds are not in the environment. This is different from phonological WM as these sounds cannot be assigned a semantic label.

Recent fMRI study on AWM in humans showed a network of activation in auditory cortex, hippocampus, and inferior frontal gyrus. They proposed a system for AWM where sound specific representations in auditory cortex are kept active by projections from hippocampus and inferior frontal cortex.

Unlike fMRI, in MEG direct neuronal activation is measured as tiny (femto Tesla) rapidly changing magnetic fields that are proportional to post synaptic dendritic currents in those pyramidal neurons that are tangentially oriented to the scalp and fire synchronously with other such neurons that are parallel to them within a cortical column. This non-invasive neurophysiological method has excellent milli-second temporal resolution of neuronal activity and good spatial resolution (better than EEG).

My MEG project aims to understand the spectro-temporal dynamics underlying this proposed system. What mechanisms underlie neural activity during retention? What is the role of hippocampus in AWM? 

Visual summary-1

Here is a visual summary of the first project 

In my experiment I contrasted working memory for pitch of a tone against working memory for spacing of a visual sinusoidal grating. Source localisation of induced activity during first second of maintenance of auditory information against silent pre-stimulus baseline showed medial prefrontal theta enhancement, cerebellar beta enhancement, and auditory cortex alpha suppression, left supramarginal gyrus theta and beta suppression. Further, I found theta phase coupling between medial prefrontal and left posterior hippocampus and right inferior frontal gyrus in addition to beta phase coupling between cerebellum and left inferior frontal gyrus (Broca's area) which was correlated with subject performance. So I speculate that, representations of sounds are kept active in the auditory cortex through covert rehearsal by Broca's area in tandem with Cerebellum and consolidated by the medial prefrontal - Hippocampus network. 

Visual summary-2

Here is a visual summary of the second project 

In my experiment where pitch of one of two tones is required to be retained for 12 s, the neuro-magnetic response showed excitation throughout the maintenance phase when compared to silent pre-stimulus baseline which is consistent with existing literature. However, when compared to a control condition that required no memorisation, the response during maintenance was not sustained over the entire duration but instead decayed to control after initial excitation. Source localisation of the evoked response during maintenance against silent pre-stimulus baseline showed activity in the auditory cortex similar to that seen for encoding phase. Similarly, source localisation of the induced response during the first second of maintenance against pre-stimulus baseline showed suppressed alpha oscillations in the auditory cortex, enhanced theta oscillations in the medial prefrontal, enhanced beta in cerebellum, suppressed beta in left supramarginal gyrus. Further I observed enhanced theta phase coupling between medial prefrontal and left anterior STG in the temporal pole . So I speculate that, for the retention of a single tone in memory, representations of the acoustic stimulus are maintained by the activation of the auditory cortex at the start of the retention phase but it is not persistent throughout the delay. Further this activity in the auditory cortex is consolidated by the medial prefrontal cortex and covert rehearsal by the cerebellum.

Video summary

Here is the video summary of this work


Here is a poster summary of this work


Source code

Here is the source code to conduct this experiment

Pradeep D., et al. (2020, September 28). MATLAB scripts to conduct Auditory Working Memory experiment.