Questions: Working Memory: Prefrontal-Parietal Neural Mechanisms

Question 1

A participant must hold two phone numbers in working memory simultaneously. One contains the digits 2, 4, 6, 8 and the other contains 1, 3, 5, 7. Compared to holding two completely unrelated number sequences, performance is likely to be:

• A. Better, because the similar sequences activate more neurons and strengthen representations
• B. The same, because working memory capacity is determined by item count, not item content
• C. Worse, because similar items recruit overlapping neural populations, causing representational interference
• D. Better, because the interleaved pattern creates a mnemonic structure the brain can exploit

Answer: (C)

Similar items recruit overlapping populations of neurons in prefrontal-parietal networks. When two patterns overlap substantially, they interfere with each other — each pattern partially corrupts the other, leading to blending or loss of items from the maintained set. This is why memorizing two phone numbers that share digits is harder than memorizing two completely unrelated sequences. Option B reflects the common 'slot' misconception — that capacity is purely about item count, independent of similarity.

Question 2

In the Fuster and Goldman-Rakic delayed-response paradigm, neurons in the dorsolateral prefrontal cortex show sustained firing during the delay period. What does this sustained firing represent?

• A. Preparatory motor activity for the upcoming response
• B. Residual sensory activation from the cue stimulus
• C. Active maintenance of the stimulus information in the absence of the original input
• D. Inhibition of competing memories from long-term storage

Answer: (C)

The key finding is that this firing persists after the cue is gone and before the response is required — it cannot be explained as a sensory response or motor preparation alone. The neuron continues to represent the remembered location throughout the blank delay. Spatial selectivity (different neurons fire for different remembered locations) and the correlation between firing stability and response accuracy confirm that this activity is the neural substrate of the memory itself, not an epiphenomenon.

Question 3

Working memory's ~3–4 item capacity limit reflects a fixed number of discrete storage 'slots' in the prefrontal cortex.

• T. True
• F. False

Answer: False

The capacity limit is better understood as an emergent consequence of representational interference, not a fixed slot count. Computational models show that recurrent prefrontal circuits can maintain a small number of distinct activity patterns simultaneously, but adding more patterns causes them to interfere and corrupt each other. The practical limit of ~3–4 items arises naturally from the network's interference dynamics. The slot metaphor is useful pedagogically but misleadingly implies that adding a 5th item suddenly fails while 4 items are perfectly maintained — in reality, degradation is gradual and depends on item similarity.

Question 4

Parietal cortex contributes to working memory by providing stable, sensory-derived representations of items, complementing prefrontal cortex's role in maintenance and goal-directed updating.

• T. True
• F. False

Answer: True

The prefrontal-parietal distinction in working memory is well-supported: parietal cortex (especially intraparietal sulcus) maintains closer ties to the sensory properties of remembered items, while prefrontal neurons show more flexible, task-dependent tuning that shifts with current goals. Damage to either region impairs working memory, but differently. The two regions form a circuit — parietal cortex provides the representational content; prefrontal cortex maintains it against interference and coordinates its use in guiding behavior.

Question 5

Why does holding similar items in working memory lead to worse performance than holding dissimilar items, even when the total number of items is the same?

Model answer: Similar items recruit overlapping populations of neurons in prefrontal-parietal working memory networks. When two items are represented by partially overlapping activity patterns, each pattern's sustained firing interferes with the other — the activation of neurons for item A partially activates the pattern for item B (and vice versa), causing the representations to blend or degrade. Dissimilar items activate more distinct neural populations, allowing them to coexist as separate stable states with minimal mutual corruption.

This interference account explains several classic findings: why recall of a word list degrades more when words are semantically similar, why articulatory similarity impairs the phonological loop, and why visuospatial items that share features are harder to hold simultaneously. It also predicts that increasing the distinctiveness of representations — through chunking, elaborative encoding, or spreading activation across more distinct neural territory — should partially counteract interference effects.