Object recognition
View an image and choose between two object labels.
Summer School 2026
Student tutorial hub
Build experiments, analyze behavior, and test artificial neural networks.
Welcome to the CVR Summer School 2026 tutorial site. Use this page to move through the workshop materials, run browser-based psychophysics tasks, launch notebooks in Google Colab, and find the example images and data used throughout the tutorials.
Suggested route through the tutorial
Follow this path during the workshop unless your instructor directs you to a specific section.
1. Read the overview
Start with the psychophysics tutorial to understand the tasks and the structure of the experiment files.
2. Try an experiment
Run one jsPsych task in your browser and download the CSV file produced at the end.
3. Clean the data
Open the matching extraction notebook in Colab and convert raw task output into tidy data.
4. Test a model
Use an ANN notebook to generate model responses for the same kind of task.
5. Compare results
Use the comparison notebook to relate human behavioral patterns to model outputs.
Psychophysics experiments
These tasks run directly in your browser. At the end of a task, save the downloaded CSV file.
Drawing recognition
View a drawing and choose between two possible labels.
Image ratings
Rate each image on a continuous slider.
N-back
Decide whether each image is novel or repeated.
Psychophysics tutorial
Learn how the experiments are organized, how stimulus metadata are used, and how saved CSV files are interpreted.
Task templates
Use the metadata examples when creating or editing stimulus lists for 2AFC, rating, or N-back experiments.
Artificial neural network tutorials
Use the same stimuli and task logic to test models like experimental subjects.
ANN tutorial
Start here for the model workflow: load images, run a model, collect outputs, and compare against behavior.
Example model results
Open example model output CSV files that can be compared with human behavioral data.
Class project: performance, reliability, and noise
Use the performance data collected from the tutorial tasks to ask how reliable behavioral measurements are, and why noise matters when comparing humans and models.
Reliability from task data
Students ran the four browser tasks. The resulting class-level CSV files can now be analyzed to estimate split-half reliability and decide which measurements are stable enough to interpret.
Noise correction tutorial
A companion notebook shows how measurement noise can reduce observed correlations, how split-half reliability is estimated, and why collecting more repetitions improves reliability.
Before opening Colab: add the CVR folder to My Drive
Do this once before starting the notebook exercises so Colab can find the shared tutorial files.
Connect the shared tutorial folder
Before running any Google Colab notebook, sign into your Google account and add the shared CVR tutorial folder to your own My Drive. This makes the tutorial images, notebooks, and data easy to access from Colab.
How to add the folder to My Drive
Why this matters: the notebooks use files stored in the shared CVR folder. Adding the folder to My Drive before opening Colab reduces path problems and keeps the tutorial workflow smoother.
Colab notebook launcher
Click a button to open the notebook in Google Colab. Sign into Google so you can save your own copy.
Tip: in Colab, use File → Save a copy in Drive before making changes. This keeps your work separate from the shared tutorial version.
| Topic | Notebook | Open | Use it for |
|---|---|---|---|
| Stimulus manipulation | Shared Colab notebook | Open in Colab | Create altered image versions such as blurred, noisy, brighter, or contrast-changed stimuli. |
| Noise correction and reliability | Shared Colab notebook | Open in Colab | Learn how measurement noise lowers observed correlations and how reliability can be used to correct them. |
| Reliability from class tasks | Shared Colab notebook | Open in Colab | Analyze the class performance CSVs and estimate split-half reliability for the tutorial tasks. |
| ANN basics | Shared Colab notebook | Open in Colab | Introductory model demonstration. |
| Facial emotion | Shared Colab notebook | Open in Colab | Model testing for facial emotion stimuli. |
| Object 2AFC model | Shared Colab notebook | Open in Colab | Generate model choices for the object 2AFC task. |
| Drawing 2AFC model | Shared Colab notebook | Open in Colab | Generate model choices for the drawing 2AFC task. |
| Ratings model | Shared Colab notebook | Open in Colab | Generate model scores for image ratings. |
| N-back model | Shared Colab notebook | Open in Colab | Generate model scores for the N-back/memorability task. |
| Extract 2AFC data | Shared Colab notebook | Open in Colab | Clean jsPsych 2AFC CSV files. |
| Extract rating data | Shared Colab notebook | Open in Colab | Clean jsPsych rating CSV files. |
| Extract N-back data | Shared Colab notebook | Open in Colab | Clean jsPsych N-back CSV files. |
| Make 2AFC task lines | Shared Colab notebook | Open in Colab | Turn metadata into editable task arrays. |
| Make rating task lines | Shared Colab notebook | Open in Colab | Turn rating metadata into editable task arrays. |
| Make N-back task lines | Shared Colab notebook | Open in Colab | Turn N-back metadata into editable task arrays. |
| Human vs ANN comparison | Shared Colab notebook | Open in Colab | Compare human behavioral results to model outputs. |
Edit and remix the tasks
Use the temporary browser playground when the tutorial asks you to change an experiment. Your edits stay in your own browser and do not change the official tutorial site.
Open the task playground
Edit a task HTML file, upload temporary images, preview the modified task, and download your edited HTML if you want to keep it.
Everything runs in the browser
The playground is only for temporary testing during the tutorial. Refreshing or closing the page resets your edits unless you download them.
Recommended workshop flow: open the playground, choose a task, edit the HTML, upload any temporary images, run the preview, then save your downloaded data or edited HTML before closing the page.
Useful resources
Quick access to the files students will use most often.
Experiments
Browser tasks
Run the object, drawing, rating, and memory tasks directly from the links above.
Notebooks
Analysis and model notebooks
Use the Colab launcher to open notebooks without installing Python locally.
Human data
Example behavioral CSVs
Sample data files for practicing extraction and analysis.
Model data
Example ANN CSVs
Model output files used in the human-vs-ANN comparison workflow.
Playground
Edit task files
Open the temporary task playground for changing HTML files, adding images, and previewing your edits.
Project
Reliability and noise
Use the class task data and noise tutorial to explore reliability, split-half consistency, and noise-corrected interpretation.
Help
Stuck?
Ask an instructor or teaching assistant. Include the page, task, or notebook name you were using and any error message you saw.