OriHendler/Decision-Making-Dynamics-Mask-the-True-Psychophysical-Capacity-of-Archerfish: formal release V1

Decision-Making Dynamics Mask the True Psychophysical Capacity of Archerfish This repository contains the data and MATLAB code to reproduce the figures in the study. All analyses run from a single data workbook. Repository layout . ├── README.md ├── code/ MATLAB scripts │ ├── ACCURACY_analysis_single_excelV5.m Figure 1 │ ├── DynamicGLM_Hendler_2pos_7models.m Figure 2 (fish 1-5) │ ├── DynamicGLM_Hendler_4quad_5models.m Figure 3 (fish 6-14) │ ├── DynamicGLM_Hendler_plot_results.m per-fish model plots + Figure 4 model comparison │ ├── ProbeVsReinforcement_Figure4.m Figure 4 probe vs reinforced scatter │ └── FISH_TRIAL_DETAILS.m per-fish trial-count table (supplementary) ├── data/ │ └── Fish_Data_Updated_with_2ACT2P_V2.xlsx └── output/ created on first run ├── FIGURE 1/ ├── FIGURE 2/ ├── FIGURE 3/ └── FIGURE 4/ Keep the three folders side by side. Each script finds the workbook in ../data and writes its results into ../output. How to run Run the scripts from the code folder in this order. Steps 2 and 3 must precede step 4, because the plotting script reads the per-fish model files they produce. ACCURACY_analysis_single_excelV5.m writes the per-fish accuracy bar plots to output/FIGURE 1. DynamicGLM_Hendler_2pos_7models.m fits the 2-position task (fish 1 to 5) and writes per-fish model files to output/FIGURE 2. DynamicGLM_Hendler_4quad_5models.m fits the 4-quadrant task (fish 6 to 14) and writes per-fish model files to output/FIGURE 3. DynamicGLM_Hendler_plot_results.m draws the per-fish model plots and the model comparison. Run it twice: FIGURE_NUM = 2, FISH_LIST = 1:5 for the 2-position fish. FIGURE_NUM = 3, FISH_LIST = 6:14 for the 4-quadrant fish. The cross-fish model comparison is written to output/FIGURE 4. ProbeVsReinforcement_Figure4.m writes the probe vs reinforced accuracy scatter to output/FIGURE 4. FISH_TRIAL_DETAILS.m reports per-fish trial counts and can be run at any time. It prints to the console and, if SAVE_TABLE = true, writes a table to output. Figure to script map Figure Script Output 1 ACCURACY_analysis_single_excelV5.m output/FIGURE 1 2 DynamicGLM_Hendler_2pos_7models.m, then DynamicGLM_Hendler_plot_results.m (FIGURE_NUM = 2) output/FIGURE 2 3 DynamicGLM_Hendler_4quad_5models.m, then DynamicGLM_Hendler_plot_results.m (FIGURE_NUM = 3) output/FIGURE 3 4 DynamicGLM_Hendler_plot_results.m (model comparison) and ProbeVsReinforcement_Figure4.m (probe vs reinforced scatter) output/FIGURE 4 Data Fish_Data_Updated_with_2ACT2P_V2.xlsx has one sheet per fish, named 1 through 14. Each row is a trial. Columns: Column Meaning slide trial index within the sheet results trial outcome. 1 = chose the target (correct), 0 = chose the non-target (incorrect). Non-numeric entries (blank, NAN, or letter codes such as M, S, A, C) mark non-response or excluded trials and are dropped by every analysis. relevance trial inclusion flag Session type Reinforcement or Probe Block name session or block identifier non_target position of the non-target stimulus on that trial (quadrant 1 to 4, or position 3/4 in the 2-position task) target position of the target, that is rewarded, stimulus on that trial recording recording identifier The model scripts read results, target, and non_target. The accuracy and trial-count scripts read results. The probe versus reinforced scatter reads results and Session type. Fish groups: Fish 1 to 5: two-position (2ACT2P) task, analyzed with the 7-model script. Fish 6 to 11: shape discrimination task, analyzed with the 5-model script. Fish 12 to 14: abstract shape task, analyzed with the 5-model script. In the probe versus reinforced analysis, fish 6 has no probe trials and is therefore excluded, leaving 13 fish. The shaded band in that figure marks the +/- 0.15 equivalence bound used in the equivalence test. Requirements MATLAB, tested with R2021a and later. exportgraphics requires R2020a or later. The scripts use base MATLAB functions (readcell, readtable, bar, errorbar, scatter, exportgraphics). The model fitting is self-contained. If MATLAB reports a missing function, the most likely dependency is the Statistics and Machine Learning Toolbox. No internet connection is required. Figures are written in .fig, .jpg, .tif, and .pdf at 600 dpi. License Data license CC-BY-4.0. How to cite Cite associated paper please. Contact hendlero@post.bgu.ac.il ori.hendler21@gmail.com Ori Hendler. Ben Gurion University of the Negev, Beer Sheba, Israel.