When searching for a target briefly presented among distractors how do people combine information across display locations to make a decision and how does the quality of the evidence entering the decision process vary with the type of items in the display? Research on accuracy in near-threshold visual search has had difficulty in distinguishing between models that make similar predictions about accuracy but make different assumptions about the underlying psychological processes. We used the diffusion model to analyse response times and accuracy data from four near-threshold search tasks which showed striking asymmetries between response-time distributions on target-present and target-absent trials. We found that performance was better explained by a model in which evidence was accumulated in parallel about each stimulus separately than one in which the evidence was pooled into a single decision process. We found that as contrast increased, the quality of the evidence entering the decision process about targets was markedly stronger than the evidence about nontargets. The overall pattern of evidence strength for stimuli on target-present and target-absent trials was consistent with a fixed-capacity memory system in which early visual processes assigned resources preferentially to targets over nontargets. The asymmetry was somewhat reduced in a letter-digit discrimination task that used heterogeneous targets and distractors, likely because heterogeneity reduces the efficiency of the preattentive filtering processes.

中文翻译：

---

近阈值视觉搜索中目标检测的扩散模型分析

当搜索在干扰者中简要呈现的目标时，人们如何结合显示位置的信息来做出决定，进入决策过程的证据质量如何随显示中项目的类型而变化？近阈值视觉搜索准确性的研究难以区分对准确性做出类似预测但对潜在心理过程做出不同假设的模型。我们使用扩散模型来分析来自四个近阈值搜索任务的响应时间和准确度数据，这些任务显示目标存在和目标缺失试验的响应时间分布之间存在显着的不对称性。我们发现，与将证据汇集到单个决策过程中的模型相比，模型可以更好地解释绩效，其中关于每个刺激的证据分别并行累积。我们发现，随着对比度的增加，进入有关目标的决策过程的证据质量明显强于有关非目标的证据。目标存在和目标缺失试验中刺激的总体证据强度模式与固定容量记忆系统一致，其中早期视觉过程优先将资源分配给目标而不是非目标。在使用异质目标和干扰物的字母数字辨别任务中，不对称性有所降低，这可能是因为异质性降低了前注意过滤过程的效率。