Effects of TMS on the decoding and electrophysiology of priority in working memory

This study demonstrates that single-pulse transcranial magnetic stimulation (spTMS) to the right intraparietal sulcus disrupts the active neural mechanisms of deprioritization in working memory, specifically by interfering with low-beta oscillatory dynamics to involuntarily restore the decodability of unprioritized memory items.

The Gist

The Big Picture: The Brain's "To-Do" List

Imagine your Working Memory is like a cluttered desk where you are trying to get work done. You have a few different files on the desk:

1. The Active File: The one you are currently reading and writing on (high priority).
2. The "Maybe Later" File: A file you need to keep, but you aren't looking at it right now because you're busy with the Active File. It's still on the desk, but it's pushed to the back (unprioritized).
3. The Trash File: A file you've finished with and don't need anymore (irrelevant).

The big question scientists have been asking is: What happens to the "Maybe Later" file?

• Theory A: You throw it in the trash (it disappears from your mind).
• Theory B: You keep it there, but you put a "Do Not Disturb" sign on it so it doesn't distract you (it's still there, just hidden).

This study wanted to find out exactly how the brain handles that "Do Not Disturb" sign and what happens if you accidentally knock the desk over.

---

The Experiment: The "Double-Back" Game

The researchers asked 12 people to play a memory game while wearing an EEG cap (which reads brain waves) and receiving a tiny, harmless zap of electricity to a specific part of their brain (the Right Intraparietal Sulcus, or rIPS). Think of this zap as a "brain poke."

The Game:

1. The Setup: You see two items (e.g., a face and a word).
2. The First Cue: A light tells you, "Focus on the Face!" The Face becomes the Active File. The Word becomes the "Maybe Later" File.
3. The Wait: You wait in silence.
4. The "Brain Poke" (spTMS): At a random moment, the researchers zap the brain.
5. The Twist: Sometimes, a second light appears later saying, "Actually, focus on the Word now!"

The Goal: They wanted to see if the "Brain Poke" could accidentally wake up the "Maybe Later" file (the Word) and make it visible to the brain's sensors, even though the person wasn't trying to look at it.

---

The Findings: The "Involuntary Wake-Up" Call

Here is what they discovered, translated into everyday terms:

1. The "Maybe Later" File is Still There (But Hidden)

Before the "Brain Poke," the brain's sensors couldn't "read" the Word. It looked like the file was gone. But the moment they zapped the brain, the sensors suddenly could read the Word again!

• Analogy: It's like a radio station that goes silent. You can't hear it. But if you tap the antenna (the brain poke), the static clears for a split second, and you hear the music again.
• Conclusion: The information wasn't deleted; it was just put into a "sleep mode" or "silent state." The zap woke it up.

2. The "Trash" File is Different

When the researchers zapped the brain after the second cue (when the Word was no longer needed and was now in the "Trash"), the zap did not wake it up immediately. It took a long time for the brain to "see" it again.

• Conclusion: The brain treats "Maybe Later" items differently than "Trash" items. "Maybe Later" is actively being held in a special state, not just thrown away.

3. The Secret Language: Low-Beta Waves

The researchers looked at the brain waves to see how this happened. They found that the "Brain Poke" worked specifically by messing with a specific rhythm in the brain called Low-Beta waves (a fast, buzzing rhythm).

• Analogy: Imagine the brain is a busy office with a specific hum (the Low-Beta rhythm). This hum is the "security system" keeping the "Maybe Later" file quiet. The "Brain Poke" temporarily broke the security system, causing the file to pop up.
• The Key Insight: When the "Maybe Later" item was designated, the brain shifted its rhythm in a specific way. The zap disrupted this rhythm, causing the item to reappear.

---

Why Does This Matter?

This study proves that deprioritizing (putting something on the back burner) is an active process, not just forgetting.

• Before this study: We thought the brain might just delete the "Maybe Later" info to save space.
• After this study: We know the brain actively puts a "Do Not Disturb" sign on it using specific brain waves (Low-Beta). It's like putting a file in a locked drawer rather than throwing it in the shredder.

The "Brain Poke" Effect:
The zap didn't just "add" information; it disrupted the lock. It broke the code that was keeping the item quiet, forcing the brain to accidentally "remember" something it was trying to ignore.

Summary in One Sentence

The brain keeps "unimportant" but "useful" information in a special "sleep mode" using a specific brain rhythm, and a tiny electrical zap can accidentally wake that information up, proving it was never truly gone.

Technical Summary

1. Problem and Research Question

The study addresses a fundamental question in cognitive neuroscience: How does the brain actively deprioritize information in working memory (WM) without discarding it?

• Context: Flexible behavior requires maintaining task-relevant information (Prioritized Memory Items, PMI) while holding potentially relevant but currently irrelevant information (Unprioritized Memory Items, UMI) in a "deprioritized" state.
• The Conflict: Previous research suggested UMIs might enter an "activity-silent" state (no neural firing), yet other studies suggest they remain active but in a distinct, "opposite" representational format.
• The Phenomenon: Single-pulse Transcranial Magnetic Stimulation (spTMS) applied to the posterior parietal cortex can trigger the involuntary retrieval of a UMI, restoring its decodability from EEG signals.
• The Gap: While the behavioral and decoding effects of spTMS on UMIs are known, the underlying neural mechanisms (specifically oscillatory dynamics) driving this deprioritization and subsequent reactivation remain unclear. The authors hypothesize that deprioritization is an active process mediated by specific oscillatory dynamics, likely in the low-beta band.

2. Methodology

Participants and Design

• Subjects: 12 neurologically healthy human participants (8 female).
• Tasks: Participants performed two tasks in alternating blocks:
  1. Double-Serial Retrocue (DSR): Two items are presented. A first retrocue prioritizes one item (PMI), leaving the other as a UMI. A second retrocue later prioritizes the UMI (or keeps the PMI). spTMS is delivered during the delay after the first cue (when the item is a UMI) and the second delay (when the item is an Irrelevant Memory Item, IMI).
  2. Single-Retrocue (SR): Similar to DSR but with only one cue. The uncued item is an IMI (no longer relevant).
• Stimulation: spTMS (single pulse) was delivered to the right intraparietal sulcus (rIPS2) at 90–110 V/m. The timing was unpredictable (50% probability) during delay periods to prevent anticipatory effects.
• Recording: Simultaneous 60-channel EEG (1450 Hz sampling) and MRI for neuronavigation.

Data Analysis

• Multivariate Pattern Analysis (MVPA): Logistic regression classifiers were trained to decode the presence of memory items (Face, Word, Motion) from EEG data.
  • Conditions: Decoding was tested for PMI, UMI, and IMI across different delay periods (DSR Delay 1.2, DSR Delay 2, SR Delay 1.2) with and without spTMS.
  • Frequency Bands: Analyses were conducted on broadband data and specific bands: Theta (4-8 Hz), Alpha (8-13 Hz), Low-Beta (13-20 Hz), and High-Beta (20-30 Hz).
• EEG Dynamics (Phase and Power):
  • Phase: Hilbert transform used to calculate Phase Locking Value (PLV) and mean phase angle shifts relative to cues and spTMS pulses.
  • Power: FFT used to analyze oscillatory power changes.
  • Statistics: Cluster-based permutation tests, Wilcoxon signed-rank tests, and permutation tests for phase/power comparisons.

3. Key Results

Behavioral Performance

• No significant effect: spTMS delivery did not significantly alter recognition accuracy or reaction times in either the DSR or SR tasks. This suggests the manipulation affected neural representations without disrupting overt behavior in this specific paradigm.

MVPA Decoding Results

• UMI Reactivation (DSR Delay 1.2): In the DSR task, when the item was a UMI, decoding accuracy was at chance before spTMS. However, concurrent with spTMS delivery, decodability of the UMI significantly rose above chance.
• IMI Reactivation (DSR Delay 2 / SR Delay 1.2): When the item was an IMI (no longer needed), decoding did not rise immediately. Instead, it rose significantly two timepoints (1000 ms) after the spTMS pulse.
• Frequency Specificity:
  • Alpha (8-13 Hz) & Low-Beta (13-20 Hz): These bands drove the decoding effects. The immediate reactivation of the UMI in DSR Delay 1.2 was robust in these bands.
  • Theta & High-Beta: No significant immediate reactivation of the UMI was observed in these bands.
• Timing Difference: The immediate reactivation of the UMI (concurrent with spTMS) was significantly stronger in DSR Delay 1.2 compared to the delayed reactivation of the IMI in other epochs.

EEG Phase and Power Dynamics

• Phase Shifts:
  • Cue Effects: The retrocue that designated an item as a UMI (DSR Cue 1) produced a unique phase shift in the low-beta band compared to the cue designating an IMI (SR Cue).
  • spTMS Effects: spTMS triggered significant phase shifts in the low-beta band specifically during DSR Delay 1.2 (UMI state). This effect was distinct from the phase shifts observed in other epochs.
• Power Dynamics:
  • Oscillatory power generally declined over time across all conditions and frequencies.
  • Crucially, the observed phase effects were not driven by amplitude/power changes, confirming that the mechanism is a genuine reorganization of phase dynamics rather than a general arousal or power increase.

4. Key Contributions

1. Mechanism of Deprioritization: The study provides evidence that deprioritizing an item in WM is an active neural process distinct from simply "dropping" the item. It involves a specific transformation of the neural representation, likely encoded in the low-beta oscillatory dynamics of the parietal cortex.
2. Oscillatory Basis of Priority: The authors identify low-beta (13-20 Hz) oscillations in the right intraparietal sulcus as the carrier of priority status. The disruption of these specific dynamics by spTMS leads to the involuntary retrieval of the UMI.
3. Temporal Dissociation: The study demonstrates a clear temporal dissociation between the reactivation of a deprioritized but relevant item (UMI) and a truly irrelevant item (IMI). The UMI is reactivated immediately by spTMS, whereas the IMI requires a longer latency, suggesting different underlying neural states.
4. Phase vs. Power: By isolating phase shifts from power changes, the study clarifies that the "activity-silent" state is likely a phase-based mechanism rather than a complete cessation of activity or a simple power suppression.

5. Significance

• Theoretical Impact: This work challenges the binary view of WM (active vs. silent) by proposing a priority map where items exist in different states of accessibility. It suggests that the brain uses low-beta oscillations to "tag" items as deprioritized, keeping them in a latent state that can be rapidly reactivated if needed.
• Methodological Advance: The combination of spTMS with high-temporal-resolution EEG and multivariate decoding offers a powerful tool for probing the causal mechanisms of cognitive states. The study validates the use of spTMS as a "ping" to test the accessibility of memory representations.
• Clinical/Computational Relevance: Understanding how the brain manages priority in WM is crucial for models of attention and executive function. Dysfunctions in these low-beta priority mechanisms could underlie deficits in disorders characterized by attentional control issues (e.g., ADHD, schizophrenia).

In summary, Fulvio and Postle demonstrate that the brain actively maintains unprioritized working memory items using low-beta oscillatory dynamics in the parietal cortex. Disrupting these dynamics with spTMS causes an immediate, involuntary retrieval of the item, proving that deprioritization is a distinct, active neural state rather than a passive loss of information.