What brain rhythms track holding durations in working memory?
When people had to remember how long a stimulus lasted two items back rather than one, posterior alpha rhythms weakened and temporal beta rhythms strengthened, while frontal theta did not change.
Source
Modulation of Alpha and Beta Oscillations during an n-back Task with Varying Temporal Memory Load
Study at a glance
- Design
- Human experiment — Within-subject EEG study: 1-back (low load) vs 2-back (high load) versions of a duration-matching n-back task, crossed with four stimulus durations
- N
- N=18 · 18 right-handed undergraduates, all completing both load conditions
- Population
- Healthy right-handed university undergraduates in China, aged 19-24
- Outcome
- Accuracy and reaction time; theta, alpha and beta EEG power across nine scalp regions
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Eighteen undergraduates watched a red circle appear for 100, 200, 400 or 800 ms on each trial and judged whether its duration matched the circle shown one trial back (low load) or two trials back (high load). EEG was recorded from 64 scalp electrodes, and the authors compared ongoing theta, alpha and beta power between the two load conditions across nine electrode clusters.
What they found
The 2-back task produced slower and less accurate responses at every duration, confirming the load manipulation worked. Alpha power was lower under high load, especially over posterior sites, and beta power was higher under high load over right-frontal and left and right central (temporal) clusters. Frontal midline theta was strongest over middle-frontal sites but was not affected by load, contrary to the authors' prediction. The load effects were already visible before stimulus onset, suggesting they reflect sustained memory demand rather than encoding of the new duration.
The limits
What it doesn't show
The sample is small (18 students) and homogeneous, and only two load levels were compared, so it cannot show a graded load-response relationship. Scalp EEG cannot pinpoint which brain areas generate the beta increase; the link to cortico-striatal timing circuits is inferred from other studies. The null theta result may reflect the identical visual stimulus used on every trial, which the authors offer as an untested explanation. Interpreting beta as 'maintenance' rests on the direction of the effect rather than a direct test of what beta does.
Key terms
- n-back task
- A working-memory task in which each item must be compared with the one presented n positions earlier; raising n increases load.
- Working memory
- The system that briefly holds and manipulates information needed for ongoing thinking.
- Alpha oscillations
- Brain rhythms around 7.5-12 Hz thought to reflect inhibition of task-irrelevant regions; they typically weaken when an area is engaged.
- Beta oscillations
- Faster rhythms (about 13-35 Hz) whose role in working memory is debated, with accounts proposing either active maintenance or inhibition.
- Frontal midline theta
- A 4-7 Hz rhythm over frontal midline sites often linked to ordering items and executive control in working memory.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What happened to posterior alpha power in the 2-back compared with the 1-back task?
Common questions
Why use durations instead of letters or pictures?
Most working-memory research uses items that look different; here every stimulus was the same red circle, so only its duration had to be held in mind, isolating temporal information.
Does the beta increase prove beta is for maintaining information?
No. It is consistent with the maintenance account and inconsistent with a simple inhibition account, but the study only shows a correlation between load and beta power.
Why didn't they subtract a pre-stimulus baseline?
Load differences were already present before each stimulus, so subtracting that period would have removed the very load-related activity they wanted to measure.
More on Working memory