N2pc attention classification

Name: n2pc

Category: visual neuroscience

Dataset: Kappenman2020N2pc (ErpCore2021_N2pc)

Objective: Binary classification

Split: Leave-subjects-out

Usage

neuralbench eeg n2pc

Show config.yaml

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

data:
  study:
    source:
      name: Kappenman2021ErpN2pc
    split:
      name: SklearnSplit
      split_by: subject
      valid_split_ratio: 0.2
      test_split_ratio: 0.2
      valid_random_state: 33
      test_random_state: 33
  # MOABB's SetRawAnnotations shifts every ErpCore2021_N2pc annotation by
  # interval[0] = -0.2 s, so events.start already sits at t_stim - 0.2 s.
  # start=0.0 + this shift => effective epoch [-0.2, +0.8] s rel. TRUE stim
  # onset, which covers the N2pc (~200-300 ms) over PO7/PO8.
  # Baseline [0.0, 0.2] in epoch-local coords maps to [-0.2, 0.0] s rel.
  # TRUE stim (pre-stimulus, canonical ERP baseline). For non-Kappenman
  # variants (e.g. reichert2020, shift=0), see datasets/*.yaml for
  # per-dataset `start` overrides. See
  # .cursor/skills/neuralbench-moabb-debug/REFERENCE.md.
  neuro.baseline: [0.0, 0.2]
  target:
    =replace=: true
    name: LabelEncoder
    event_types: Stimulus
    event_field: description
    return_one_hot: true
    aggregation: trigger
  trigger_event_type: Stimulus
  start: 0.0
  duration: 1.0
  summary_columns: [description]
compute_class_weights: true
brain_model_output_size: &brain_model_output_size 2
trainer_config.monitor: val/bal_acc
trainer_config.mode: max
loss:
  name: CrossEntropyLoss
  kwargs:
    label_smoothing: 0.1
metrics: !!python/object/apply:neuralbench.defaults.metrics.get_classification_metric_configs
  - *brain_model_output_size

Description

The N2pc classification task involves decoding the location of visual spatial attention from EEG recordings. The N2pc is an ERP component reflecting the deployment of visual-spatial attention to a target item in a visual search array, observable as a negative deflection contralateral to the attended item around 200-300 ms post-stimulus. We use the Kappenman2020N2pc dataset [Kappenman2020N2pc], part of the ERP CORE (Compendium of Open Resources and Experiments), which contains EEG data from 40 subjects performing a visual search task.

Additional Datasets

The following additional dataset from MOABB can also be used with this task:

• Reichert2020 – 18 subjects, covert spatial attention N2pc paradigm

To run with an alternate dataset:

neuralbench eeg n2pc --datasets reichert2020

References

[Kappenman2020N2pc]

Kappenman, E. S., et al. “ERP CORE: An open resource for human event-related potential research.” NeuroImage 225 (2021): 117465.