# Exp04 通道消融

> 本页由仓库源文件生成。请不要直接编辑本页；修改下方 source 文件后运行 `vp run docs:generate`。

> 中文支持说明：本页是中文站点镜像，保留论文标题、证据状态和来源字段的原始表述，避免未经核实的翻译改变证据边界。

> 来源: `experiments/exp04-channel-ablation/protocol.md`

## Experiment 4: Low-Channel Ablation

Status: `pre-data protocol draft`

## Goal

Evaluate whether the system can move from standard EEG cap acquisition toward lower-channel wearable configurations.

## Strategy

1. Collect full-core 16ch data.
2. Replay 12ch, 8ch, 6ch, and 3ch subsets offline.
3. Move only stable subsets into online tests.

## Channel Sets

```text
16ch full-core
12ch product-core
8ch minimal-core
6ch hybrid-minimal or SSVEP-prior
3ch SSVEP-only: O1/Oz/O2
3ch MI-only: C3/Cz/C4
```

## Metrics

* SSVEP accuracy and ITR by channel subset.
* MI accuracy and false activation by channel subset.
* Closed-loop task success and completion time.
* Setup time and user burden.

## Analysis Plan

See `../analysis-plan.md`. Complete `../pre-data-freeze-checklist.md`, exact channel-set definitions, and a session record before collecting evidence-producing trials.

Required pre-session records:

* `session_id`;
* `hardware_inventory_id`;
* `threshold_freeze_record_id`;
* protocol version and software commit;
* full montage and reduced channel-set definitions;
* preprocessing and decoder IDs;
* fixed-model versus subset-retrained analysis mode;
* training/pilot/test split;
* randomization seed/file and block order.

Trial inclusion rule:

* Use the same attempted-trial log as the source full-channel condition.
* Keep channel-masked offline replay separate from subset-retrained analysis.
* Do not promote a reduced channel set to online control unless progression thresholds were frozen before testing.

Primary endpoint:

* performance retained by reduced channel sets relative to the 16-channel reference.

Baseline interpretation:

* 16ch full-core is the reference acquisition condition;
* 12ch, 8ch, and 6ch subsets test staged low-channel reduction;
* 3ch SSVEP-only and 3ch MI-only subsets test modality-specific minimal layouts.

Online-control rule:

* reduced channel sets should move from offline replay to online control only after pre-set retention and false-activation criteria are met. Numeric thresholds and exact electrode sets are `needs confirmation`.
* online progression decisions must cite threshold IDs from `experiments/templates/threshold-freeze-record.md`.

***

> 来源: `experiments/exp04-channel-ablation/metrics.md`

## Metrics

| Metric | Definition | Status |
| --- | --- | --- |
| SSVEP accuracy | correct accepted selections / attempted SSVEP trials by channel set | definition draft frozen |
| SSVEP ITR | ITR by channel set using `../analysis-plan.md` raw-window and practical-selection variants | definition draft frozen; final `T` values need session binding |
| MI accuracy | correct mode commands / attempted MI trials by channel set | definition draft frozen |
| false activation | no-control accepted commands / no-control trials by channel set | definition draft frozen |
| task success | closed-loop success / attempted closed-loop trials by channel set | definition draft frozen; online use needs progression gate |
| setup burden | setup time and subjective burden linked by session/questionnaire ID | protocol choice needs confirmation |

***

> 来源: `experiments/exp04-channel-ablation/variables.md`

## Variables

| Factor | Levels | Notes |
| --- | --- | --- |
| channel set | 16 / 12 / 8 / 6 / 3 | define exact electrodes before acquisition |
| task module | SSVEP / MI / closed-loop | evaluate module and system levels |
| analysis mode | offline replay / online control | offline first |
| participant | within-subject | preferred for ablation |

## Freeze Requirements

* Exact electrode labels for 16/12/8/6/3 channel sets must be frozen before acquisition or replay.
* Fixed-model channel masking and subset-retrained analyses must be labeled separately.
* Online reduced-channel control requires a threshold-freeze record and progression decision.
* Report reduced-channel performance relative to the 16-channel reference, not only as absolute accuracy.