Cambridge Bicycle Memory Test¶

This pre-configured paradigm lets you collect data using the Cambridge Bicycle Memory Test - a variant of the Cambridge Face Memory Test (CFMT) that uses bicycles as stimuli instead of faces. With this "preset", you can quickly assess object recognition abilities and determine whether face recognition deficits are specific to faces or extend to other object categories. This page explains how to use the preset, what data you'll collect, and how to analyze your results.

By comparing performance on face memory tests with performance on bicycle memory tests, researchers can distinguish between face-specific and domain-general visual processing deficits.

Example Cambridge Bicycle Memory Test stimulus — Example stimulus from the Cambridge Bicycle Memory Test showing bicycles that participants must learn to recognize and discriminate.

Quick Start

Create or open an experiment from your Dashboard
Click Add pre-configured tasks
Choose "Cambridge Bicycle Memory Test"
Preview your experiment

New to Meadows? See the Getting Started guide for a complete walkthrough.

Background¶

The Cambridge Bicycle Memory Test was developed by Dalrymple and colleagues as part of a battery of tests to evaluate visual recognition abilities across different stimulus categories¹². The test follows the identical structure and procedure as the CFMT, ensuring that any performance differences between faces and bicycles can be attributed to category-specific processing rather than task demands.

Research using these Cambridge Memory Tests has shown that while some individuals with prosopagnosia show face-specific deficits, others demonstrate broader object recognition impairments². The bicycle test is particularly useful because bicycles, like faces, require fine-grained discrimination between visually similar exemplars.

Test Structure¶

The preset includes a single task (cmt_bikes) that follows the standard Cambridge Memory Test format. Participants typically complete the test in 10-15 minutes.

Practice Phase - Familiarization with the task
Introduction Phase - Learn target bicycle identities (shown from 3 viewpoints each)
Review Phase - Review all target bicycles
Novel Test - Identify targets among novel distractor bicycles
Noise Test - Identify targets with visual noise added (more difficult)

In each test trial, participants view 3 bicycles and press 1, 2, or 3 to indicate which bicycle is one of the target identities.

Tasks Included¶

This preset contains the following task:

Task Name	Task Type	Description
`cmt_bikes`	`cmt`	The bicycle memory test with all phases

Data¶

For general information about the various structures and file formats that you can download for your data see Downloads.

Annotations

As stimulus-wise "annotations" (table rows), with columns:

trial - numerical index of the trial
time_trial_start - timestamp when the test images were displayed (seconds since 1/1/1970)
time_trial_response - timestamp when the participant responded (seconds since 1/1/1970)
stim1_id - Meadows internal id of the stimulus
stim1_name - filename encoding trial information
label - the key pressed by the participant (1, 2, or 3)

Analysis¶

Scoring¶

The primary outcome is the total score (number of correct responses). Score calculation is identical to the CFMT.

SpreadsheetPythonR

In Google Sheets or Microsoft Excel, you can score the test:

Load Data: Download the annotations data in Table format from Meadows, then import the Meadows_myExperiment_v1_cmt_bikes1_annotations.csv file into your spreadsheet application.
Create Correct Column: Add a formula to check if the response was correct. In a new column (e.g., column D if label is in column C):
```
=IF(C2="1", 1, 0)
```
This assumes the correct answer is always "1" (the target in the first position). Drag this formula down for all rows.
Calculate Total Score: Sum the correct responses:
```
=SUM(D:D)
```
Calculate Percentage: Divide by the total number of trials:
```
=SUM(D:D)/COUNTA(C:C)*100
```
Calculate Reaction Time: Create a formula to calculate RT in milliseconds:
```
=(time_trial_response - time_trial_start) * 1000
```
Then calculate the mean RT across all trials.

import pandas as pd

# Load the annotations data
df = pd.read_csv('Meadows_myExperiment_v1_cmt_bikes1_annotations.csv')

# For standard CMT, the target is typically in position 1
df['correct'] = df['label'] == '1'

# Calculate total score
total_score = df['correct'].sum()
max_score = len(df)
percentage = (total_score / max_score) * 100

print(f"Total Score: {total_score}/{max_score} ({percentage:.1f}%)")

# Calculate reaction time
df['rt_ms'] = (df['time_trial_response'] - df['time_trial_start']) * 1000
print(f"Mean RT: {df['rt_ms'].mean():.0f} ms")

library(tidyverse)

# Load the annotations data
df <- read_csv('Meadows_myExperiment_v1_cmt_bikes1_annotations.csv')

# Calculate accuracy
df <- df %>%
  mutate(
    correct = label == '1',
    rt_ms = (time_trial_response - time_trial_start) * 1000
  )

# Calculate total score
results <- df %>%
  summarise(
    total_score = sum(correct),
    max_score = n(),
    percentage = mean(correct) * 100,
    mean_rt = mean(rt_ms)
  )

print(results)

Interpreting Scores¶

Performance on the Cambridge Bicycle Memory Test is typically compared with performance on the CFMT to assess the specificity of recognition deficits. Individuals with face-specific processing deficits should show relatively preserved bicycle recognition, while those with broader object recognition problems may show deficits on both tests.

Individual Differences

Performance on object recognition tests can vary across individuals. When comparing with face recognition, consider that some people may show category-specific processing differences.

References¶

Dalrymple, K.A., Elison, J.T., & Duchaine, B. (2017). Face-selective and domain-general visual processing deficits in children with developmental prosopagnosia. Quarterly Journal of Experimental Psychology, 70(2), 259–275. doi:10.1080/17470218.2015.1122642 ↩
Barton, J. J., Albonico, A., Susilo, T., Duchaine, B., & Corrow, S. L. (2019). Object recognition in acquired and developmental prosopagnosia. Cognitive Neuropsychology, 36(1-2), 54–84. doi:10.1080/02643294.2019.1593821 ↩↩