Skip to main content
Jianshu Wang
Back to portfolio

Community environments and everyday well-being

How are community environments associated with residents’ well-being and the way they evaluate their own community?

Manuscript submitted

Institution
Lifedu Educational Technology Co., Ltd.
Supervision
Dr. Ze Geng, Principal Investigator

I preprocessed electroencephalography (EEG) recordings, extracted centroparietal positivity (CPP) slopes, and fitted and compared behavioral models for this collaborative study.

Task illustration: a participant wearing an EEG cap uses a keyboard to classify a community image on a computer display.Opens in a new tab.
Task illustration. Electroencephalography (EEG) was recorded during classification.
  1. Illustrative residential community with apartment buildings, trees and a path.
    Community imageClassify and respond
  2. pleasant
    Then an affective wordClassify and respond

Mapping 1

FOwn communityPositive
JOther communityNegative

Mapping 2

FOwn communityNegative
JOther communityPositive
One stimulus at a time. “Own” refers to the participant’s residential community. Example mappings; category-to-key assignments were counterbalanced.

From recordings to a decision model

Trial-level CPP slope

Derived from centroparietal EEG

Time
Schematic EEGFitted slope in a selected interval

CPP included in the drift-rate specification

Evidence accumulation

ChoiceResponse time

Observations for model fitting

Correct responseError responseTime →
DecisionDrift rate

Schematic accumulation paths

Matched behavioral and EEG trials informed the model; CPP slope entered its drift-rate specification. Illustrative evidence paths reach a correct or error boundary to produce a decision; these are not study results.

I retained valid error responses for decision modeling.

Model fitting and comparison

CPP slopes were extracted from trial-level EEG averaged across Pz, Cz, CP1 and CP2. Models were fitted using aggregate maximum-likelihood estimation in PyDDM and compared using BIC. CPP slope was a measured covariate; drift rate was a model parameter. The model separated accumulation from nondecision time, so boundary-crossing time alone is not the full response time.

Error trials were excluded from response-time and D-score summaries, while valid error choices were retained for decision modeling.

Community evaluations research overview

Further study details and results will be added after publication.