Stimulus Data

The LAION-fMRI dataset includes comprehensive visual stimuli presented to participants during fMRI scanning.

Stimulus Sets

The dataset may include multiple stimulus sets for different experiments:

stimuli/
├── localizer/     # Functional localizer stimuli
└── n_back/        # N-back task stimuli

Each set includes its own metadata file describing the specific stimuli and their properties.

Visual Stimuli

Stimulus Categories

The dataset includes a variety of visual stimuli:

• Natural images: Photographs of natural scenes

• Object categories: Various object types and categories

• Scene categories: Indoor and outdoor scenes

• Face stimuli: Human faces with varying expressions and identities

Stimulus Format

Technical Specifications:

• Image files: PNG/JPEG format

• Resolution: 1024x1024 pixels

• Color space: RGB

• Bit depth: 8-bit per channel

Stimulus Metadata

File Organization

Stimulus files are organized as follows:

stimuli/
├── localizer/       # Functional localizer stimuli
│   ├── images/      # Stimulus image files
│   ├── category_01/
│   ├── category_02/
│   └── ...
│   ├── stimuli.tsv  # Stimuli list and reference
│   ├── stimuli.json # Stimuli metadata
│   └── README.txt   # Stimulus documentation
└── n_back/          # N-back task stimuli

Stimuli list and Reference

The stimuli.tsv file contains a list of all stimulus images along with their feature values, identifiers and categories.

stimulus_id category    filename    luminance   contrast    experiment_id    duration
stim_0001   face    face_001.png    127.5   0.42    exp_01  500
stim_0002   scene   scene_001.png   115.2   0.38    exp_01  500
stim_0003   object  object_001.png  132.1   0.45    exp_02  750

Metadata Fields

Each stimulus includes comprehensive metadata:

Primary Fields:

• stimulus_id: Unique identifier for each stimulus

• category: Primary category label

• subcategory: Fine-grained category (if applicable)

• filename: Image filename

• presentation_order: Order of presentation in experiment

Visual Properties:

• luminance: Mean luminance value

• contrast: RMS contrast

• spatial_frequency: Dominant spatial frequency

• color_histogram: RGB color distribution

Experimental Information:

• experiment_id: Which experiment used this stimulus

• condition: Experimental condition

• trial_type: Type of trial (e.g., target, distractor)

• duration: Presentation duration in milliseconds

Stimuli Metadata

The stimuli.json file contains detailed metadata for the stimulus set:

{
    "Name": "LAION-fMRI Localizer Stimuli",
    "BIDSVersion": "1.8.0",
    "License": "CC-BY-4.0",
    "Authors": [
        "ViCCo-Group"
    ],
    "Description": "Visual stimuli used in the LAION-fMRI functional localizer experiment",
    "StimulusPresentation": {
        "ScreenSize": [1920, 1080],
        "ScreenDistance": 60,
        "ScreenRefreshRate": 60
    },
    "ImageProperties": {
        "Resolution": [1024, 1024],
        "ColorSpace": "RGB",
        "BitDepth": 8,
        "Format": "PNG"
    },
    "Categories": [
        {
            "CategoryID": "faces",
            "CategoryName": "Human Faces",
            "NumberOfStimuli": 120,
            "Subcategories": ["neutral", "happy", "sad"]
        },
        {
            "CategoryID": "scenes",
            "CategoryName": "Natural Scenes",
            "NumberOfStimuli": 120,
            "Subcategories": ["indoor", "outdoor"]
        },
        {
            "CategoryID": "objects",
            "CategoryName": "Objects",
            "NumberOfStimuli": 120,
            "Subcategories": ["tools", "furniture", "vehicles"]
        }
    ],
    "VisualProperties": {
        "MeanLuminance": 127.5,
        "ContrastRange": [0.2, 0.6],
        "SpatialFrequencyRange": [0.5, 8.0]
    }
}

Distribution of Stimuli

Distribution of stimuli across categories in the LAION-fMRI dataset.

Loading Stimulus Data

Using the Python Package

The laion-fmri-dataloader package provides convenient functions for loading stimulus data:

from laion_fmri_dataloader import LAIONfMRIDataset
import matplotlib.pyplot as plt
from PIL import Image

# Initialize dataset
dataset = LAIONfMRIDataset(data_dir='./laion-fmri-data')

# Load stimulus metadata
stimuli_metadata = dataset.load_stimuli(stimulus_set='localizer')
print(f"Total stimuli: {len(stimuli_metadata)}")
print(f"Categories: {stimuli_metadata['category'].unique()}")

# Load a specific stimulus image
stimulus_img = dataset.load_stimulus_image(
    stimulus_set='localizer',
    stimulus_id='stim_0001'
)

# Display the image
plt.figure(figsize=(8, 8))
plt.imshow(stimulus_img)
plt.title(f"Stimulus: {stimuli_metadata.loc[0, 'stimulus_id']}")
plt.axis('off')
plt.show()

# Get stimuli by category
face_stimuli = dataset.get_stimuli_by_category(
    stimulus_set='localizer',
    category='faces'
)
print(f"Face stimuli: {len(face_stimuli)}")

# Load multiple images at once
images = dataset.load_stimulus_images(
    stimulus_set='localizer',
    stimulus_ids=['stim_0001', 'stim_0002', 'stim_0003']
)

Filtering Stimuli

# Get all face stimuli
face_stimuli = stimuli[stimuli['category'] == 'face']

# Get stimuli from specific experiment
exp1_stimuli = stimuli[stimuli['experiment_id'] == 'exp_01']

# Filter by visual properties
high_contrast = stimuli[stimuli['contrast'] > 0.4]

Stimulus Validation

Quality Checks

All stimuli have undergone quality control:

• Resolution verification: All images meet minimum resolution requirements

• Color space validation: Correct RGB color space

• Duplicate detection: No duplicate images

• Metadata completeness: All required fields populated