PsyNet如何基于analyze_recording的analysis结果展示试次自定义反馈

实现方案 #

1. 配置单试次分析逻辑 #

首先要为你的试次类实现analyse_response方法，确保每个试次完成后会生成敲击次数的分析数据：

class YourTrial(Trial):
    # 保留你原有Trial类的其他配置
    def analyse_response(self, response, participant):
        # 从用户提交的响应数据中统计当前试次的敲击次数，根据你实际的响应字段调整
        tap_count = len(response.get("tap_events", []))
        return {
            "num_resp_raw": tap_count
        }

2. 两种方式获取总敲击次数 #

方式一：直接在performance_check中遍历汇总（无需额外配置）

不需要等待框架生成聚合分析结果，直接遍历当前试次块下的所有试次手动汇总：

def performance_check(self, experiment, participant, participant_trials):
    n_taps_detected = 0
    for trial in participant_trials.trials:
        if trial.status == "completed" and trial.analysis is not None:
            n_taps_detected += trial.analysis.get("num_resp_raw", 0)
    failed = participant_trials.failed
    return {"score": n_taps_detected, "passed": not failed}

方式二：配置试次块聚合分析（可直接调用participant_trials.analysis）

如果要使用你原来代码里的participant_trials.analysis['num_resp_raw_all']写法，需要给PracticeTrialMaker添加聚合分析方法：

class PracticeTrialMaker(StaticTrialMaker):
    # 保留你原来的配置
    give_end_feedback_passed = True
    performance_check_type = "performance"
    performance_check_threshold = 0
    end_performance_check_waits = True

    # 新增聚合分析方法，框架会自动调用生成participant_trials.analysis
    def aggregate_analysis(self, trials, participant):
        total_taps = 0
        for trial in trials:
            if trial.status == "completed" and trial.analysis:
                total_taps += trial.analysis.get("num_resp_raw", 0)
        return {"num_resp_raw_all": total_taps}
    
    # 你原来的反馈页面和性能检查逻辑可以保持不变
    def get_end_feedback_passed_page(self, score):
        how_many_taps = "NA" if score is None else f"{(score):.0f}"
        return InfoPage(
            Markup(
                f"你本次敲击了 <strong>{how_many_taps} 次</strong>。"
            ),
            time_estimate=5,
        )
    
    def performance_check(self, experiment, participant, participant_trials):
        n_taps_detected = participant_trials.analysis['num_resp_raw_all']
        failed = participant_trials.failed
        return {"score": n_taps_detected, "passed": not failed}

注意事项 #

你已经配置了end_performance_check_waits = True，该配置会保证框架在所有试次的分析逻辑执行完成后再调用performance_check，不会出现analysis为空的问题。

内容的提问来源于stack exchange，提问作者vanirajendran