FEATURE 2778

/ Hippocampal Memory and Cognitive Function

This page shows the detailed analysis of a specific feature in the Sparse Autoencoder (SAE). It includes the semantic interpretation generated by the LLM, the top activating documents that trigger this feature, and statistical metrics like density and activation distribution.

Semantic Interpretation

gemma3:12b #13

The neuron appears to activate in contexts related to memory, particularly memory deficits and impairments, especially within the hippocampus and related structures. This includes face and facial expression memory, prospective memory, and memory impacted by conditions like epilepsy and meningitis. The activation is also linked to executive functions and language performance, suggesting a broader role in cognitive processes dependent on memory. The presence of brain lesions, particularly in the parietal lobe, and the impact of sleep quality further reinforce this connection. The focus is on *impaired* memory and cognitive function, not simply memory itself.

STATISTICS & DISTRIBUTION

Density

0.01000

Peak Act

3.68

0.0 Max

Global Context

TOP ACTIVATING CONTEXTS

DOC #691 ANALYZE

ACT: 3.6756

Face and facial expression memory in temporal lobe epilepsy patients: preliminary results. Right temporal lobe structures are involved in face and facial expression processing and in mnestic functions. Face and facial expression memory was investigated in 15 patients with left (LTLE) and 18 patients with right (RTLE) temporal lobe epilepsy as well…

DOC #822 ANALYZE

ACT: 2.9641

Spatial perception errors do not predict pointing errors by individuals with brain lesions. We examined the relationship between errors in sensorimotor transformations (SMT) for reaching to external targets and visual and kinesthetic spatial perception of those targets by participants with damage to the posterior parietal lobule (PPL) and adjacent…

DOC #879 ANALYZE

ACT: 2.8328

Initial planning benefits complex prospective memory at a cost. The effect of initial planning on complex prospective memory was investigated using a virtual environment and a sample of healthy young adults (N = 34). Participants were assigned to either an initial planning or a control condition and were asked to complete a series of time- and eve…

DOC #78 ANALYZE

ACT: 2.6724

Implicit semantic perception in object substitution masking. Decades of research on visual perception has uncovered many phenomena, such as binocular rivalry, backward masking, and the attentional blink, that reflect 'failures of consciousness'. Although stimuli do not reach awareness in these paradigms, there is evidence that they nevertheless un…

DOC #1 ANALYZE

ACT: 2.5290

Effect of sleep quality on memory, executive function, and language performance in patients with refractory focal epilepsy and controlled epilepsy versus healthy controls - A prospective study. We aimed to evaluate the effect of sleep quality on memory, executive function, and language performance in patients with refractory focal epilepsy and con…

TOPOLOGY

Child of Neuronal Circuitry and Cellular Processes
#23
Child of Sensory-Motor Integration
#2972

CORRELATIONS

W – Weight-space · similarity between decoder vectors (features that point in similar directions in the embedding space).

D – Data / co-activation · features that tend to fire together on the same documents (co-occurrence in the dataset).

Neuronal Circuitry and Cellular Processes 0.80
Feature 1213 0.50
Sensory-Motor Integration 0.30
Feature 706 0.20
Developmental Disparities & Intervention 0.20

Ollama Model

Temp

K+

K-

System Prompt

You are a meticulous researcher investigating a specific neuron in a language model. Your task is to determine what behavior this neuron is responsible for: what concepts, topics, or linguistic features does it activate on?

INPUT DESCRIPTION: You will receive two inputs: 1) Maximum Activation Examples and 2) Zero Activation Examples.
1. You will be given several text examples that activate the neuron, along with a number indicating how strongly it was activated. This means there is some feature, concept, or pattern in this text that 'excites' this neuron.
2. You will also be given several text examples that do NOT activate the neuron. This means the feature or concept is not present in these texts.

OUTPUT DESCRIPTION: Given the inputs provided, complete the following tasks.
1. Based on the MAXIMUM ACTIVATION EXAMPLES, list potential topics, concepts, themes, and features they have in common. Be specific. You may need to look at different levels of granularity. List as many as possible. Give greater weight to concepts more prominent in higher-activation examples.
2. Based on the zero activation examples, systematically exclude any topic/concept/feature listed above that also appears in the zero activation examples.
3. Based on the two previous steps, perform a thorough analysis of which feature, concept, or topic, at which level of granularity, is likely to activate this neuron. Use Occam's razor, as long as it fits the evidence provided. Be highly rational and analytical.
4. Based on step 3, summarize this concept in 1-8 words, in the form FINAL: <explanation>. Do NOT return anything after these 1-8 words.

Respond EXCLUSIVELY with valid JSON: {'label': '...', 'description': '...'}

FEATURE 2778

Semantic Interpretation

Statistics Explained

Global Context