Pinging Hidden Working Memory Using Brain Stimulation and Precision Imaging

Imagine hosting a dinner party. Your mind is filled with tasks like buying ingredients, cooking, and awaiting guests' arrivals. Yet you only pay attention to one task at a time— be it boiling pasta or answering the doorbell. Your ability to “actively” hold and manipulate limited information against a broader “hidden” cognitive background is known as working memory (WM), a pivotal human cognitive function. Understanding the neural basis of WM impacts a variety of fields, from neuropsychiatric treatments as in frontal lobe disorders to improving learning in children to building artificial general intelligence. Active WM has been studied extensively, and is often considered the core of WM. Despite its significance in computational models, the concept of background or “hidden” WM remains less understood. This is because hidden WM is marked by minimal neural activity, potentially hidden in short-term synaptic changes, making it difficult to see with neuroimaging. This elusive nature has led to propositions that hidden WM is crucial for optimizing brain energy use and shielding information from distractions. Non-invasive brain stimulation techniques like transcranial magnetic stimulation (TMS) can briefly reactivate these hidden states. In this study, I use an innovative and technically demanding simultaneous combination of TMS with precision brain imaging to “ping” or visualize hidden states; similar to how a ship emits echo pulses to map the ocean floor. This study aims to advance our understanding of how the brain supports WM through complementary active and hidden components.

Brain imaging studies show that active WM implicate a core brain circuit called the Multiple-demand (MD) circuit, which co-activates during various cognitive tasks. Using new precision fMRI approaches, I introduced a new model which redefines our understanding of how this circuit works during WM tasks. Instead of separate areas handling different WM processes, the new model puts a key emphasis on information exchange at the intersections of the MD circuit and surrounding specialized areas. Thus here, I will explore if connections between MD and nearby specialized circuits also support hidden WM processes.

My first aim was to develop a novel TMS-fMRI setup for high-resolution whole brain imaging. Typical current TMS-fMRI setups have limited spatial resolution for fMRI because the need to accommodate the TMS coil has limited the number of radio-frequency (RF) MR channels to 14. With support from technical and MRI specialists, I developed a novel setup incorporating two flexible RF coils that wrap around both the head and the TMS coil, resulting in a 22-channel configuration. This is a significant advance as it makes high-precision imaging possible. Pilot fMRI data already demonstrate the new setup outperforms existing state-of-the-art setups.

Next, I used the new setup to scan participants as they solved cognitive tasks involving specific visual stimuli, like faces and houses. Importantly, I administered TMS pulses to the core MD circuit during rest periods, not while the tasks were being performed. I predicted that these TMS pulses would travel through "silent" connections to the brain areas associated with each task's stimuli, even though these connections were not actively in use. Preliminary results indeed support this hypothesis, with TMS differentially modulating activity in the task-relevant brain circuits. This is proof of concept that TMS-fMRI can be used to probe hidden connections between task and stimuli, expanding on previous work suggesting that TMS can momentarily reactivate hidden WM states. This innovative approach aims to (1) build a novel neural account for how the brain toggles between foreground and background information and (2) offer unmatched resolutions for cognitive and clinical research.

by Dr. Moataz Assem