Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› psychrev-conceptual-exhibits
psychrev-conceptual-exhibits
GitHub用于设计《Psychological Review》论文中的理论图表，包括模型架构图和模型行为模拟图。旨在展示理论结构与预测结果，而非实验数据。提供设计规范、参数透明度要求及APA图注标准，确保图表具备理论说服力且可审查。
触发场景
需要绘制包含组件与信息流的模型架构图
需要生成展示模型预测行为的模拟曲线或相图
审稿人要求展示模型行为而非仅架构
需要将模型预测与现有数据或竞争模型进行对比
安装
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill psychrev-conceptual-exhibits -g -y
SKILL.md
Frontmatter
{
"name": "psychrev-conceptual-exhibits",
"description": "Use when designing the figures of a Psychological Review manuscript — model\/architecture diagrams and simulation-of-model-behavior plots (NOT experiments). Builds the exhibits; it does NOT derive the predictions they display (psychrev-argument-development) or set the model's scope (psychrev-boundary-conditions)."
}
Conceptual Exhibits: Model Diagrams & Simulation Figures (psychrev-conceptual-exhibits)
When to trigger
- The theory is built and derived; now it must be shown
- Your figure is boxes-and-arrows with no specified dynamics
- You have a model but no plot of what it does
- A reviewer will ask "show me the model's behavior, not just its architecture"
The two figure types Psychological Review actually uses
Figures here carry theoretical work. They are almost never plots of an experiment you ran (you ran none). There are two legitimate kinds:
- Model / architecture diagrams — the structure of the theory: components, representations, information flow, feedback. Every box is a defined construct or module; every arrow is a specified process or relationship from the theory, not decoration.
- Simulation-of-model-behavior figures — the model's output: predicted curves, surfaces, phase portraits, parameter sweeps, fits to existing data. This is the figure that does the persuasive work — it shows the model produces (or fails to produce) the diagnostic phenomena.
A third, supporting kind: model-vs-data and model-vs-rival overlays, where the model's prediction is plotted against published data points or against a rival model's prediction.
Design rules for model diagrams
- Every box maps to a named assumption/construct/module in the text; every arrow to a specified process. If a reader cannot read the diagram off the model description, it is decoration.
- Distinguish core structure from implementational detail visually (e.g., shading), so reviewers see what is load-bearing.
- Keep the level of analysis consistent within a diagram (do not mix computational and implementational components without saying so).
- Label representations and processes with the same terms used in the equations/algorithm.
Design rules for simulation figures
- Plot the diagnostic behaviors — the patterns that distinguish your theory from rivals.
- State, in the caption, the parameters used and whether they were fit or set a priori; a simulation figure with undisclosed parameters is unreviewable.
- Show parameter sweeps where the qualitative pattern is the prediction (the pattern should
be robust across the plotted range, per
psychrev-boundary-conditions). - Overlay existing data when confronting the model with evidence; cite the data source.
- When comparing to a rival, plot both predictions on the same axes; do not compare across panels with different scales.
- Mark which curves are predictions (model not tuned on them) vs. fits (tuned).
Caption discipline (APA)
A Psychological Review caption is self-contained: what the panel shows, the model and parameters that produced it, the data source if overlaid, and what theoretical point it makes. Numbering and labels must match the text exactly; reference every figure in the prose.
Checklist
- Every model-diagram box is a defined construct/module; every arrow a specified process
- Core structure is visually distinguished from implementational detail
- At least one simulation figure shows the model's diagnostic behavior
- Simulation captions disclose parameters and whether they were fit or set a priori
- Predictions (untuned) are visually distinguished from fits (tuned)
- Model-vs-data and model-vs-rival overlays share axes and cite data sources
- Figures meet APA resolution/format specs; numbering and labels match the text
Anti-patterns
- A box-and-arrow diagram that cannot be read off the model description (decoration)
- Presenting the architecture but never the behavior the architecture produces
- Simulation figures with undisclosed or cherry-picked parameters
- Comparing model and rival across panels with mismatched axes/scales
- Plotting a fit and implying it is a prediction
- A figure of an experiment — there is no experiment; data appear only to constrain the model
Output format
【Figure list】[Fig 1: diagram | Fig 2..n: simulation/behavior | overlays vs. data/rival]
【Diagram check】boxes=constructs, arrows=processes, core vs. impl. distinguished: yes / fix
【Behavior figures】diagnostic patterns shown; parameters disclosed (fit vs. a priori): yes / fix
【Predictions vs. fits】visually marked: yes / fix
【Captions】self-contained, APA, matched to text: yes / fix
【Next step】psychrev-contribution-framing → psychrev-writing-style
版本历史
- 1839142 当前 2026-07-05 14:15
