EAGER: Cognitive modeling of strategies for dealing with errors in mobile touch interfaces

众所周知，手机和平板电脑上的触摸界面容易出错。改善触摸屏设备可用性进展缓慢的一个合理原因是，人机交互的研究和设计工作相对狭窄地关注于识别、理解和消除人为错误，主要关注于特定的个人触摸交互，而不是人们需要使用设备完成现实世界任务的更广泛的任务知识。这个项目采取了不同的视角，表明使用错误代表了人们在适应任务和设备的复杂性方面的失败，并且许多这些问题可以追溯到界面设计师没有充分考虑到用户将需要完成任务的认知任务知识。许多触摸屏设计指南都是随意的，都是基于常规操作和用户期望。本项目旨在通过将认知任务分析和错误分析的科学原理引入移动设备触摸交互的设计和分析，改变移动设备触摸交互的研究和实践形态。这个项目可以极大地改善触屏界面的用户界面设计，从而造福社会。该项目将开发认知模型，即在某些方面表现得像人类的计算机程序。该项目将专门开发触摸交互的认知模型，供设计师在设计新的触摸屏界面时使用，以确定用户在使用新界面时可能遇到的问题。该项目包括三个阶段。第一阶段将在移动平台上开发一套小型仪表接口，为容易出错的任务收集人工数据。实验将检查序列中特定点错误概率的变化，为特定动作提供的视觉反馈，以及用户对错误恢复和避免动作的了解。第二阶段将遵循认知建模方面的其他成功努力，分析我们探索性实验中的用户数据，目标是开发我们所观察到的策略的表示。一个自动化的过程将探索一个合理的认知策略的搜索空间，以确定解释在第一阶段收集的性能数据的模型。第三阶段将验证与认知策略有关的假设，这些假设有助于在前两个阶段建立的触摸屏设备的成功和不成功的使用。通过关注专业知识和失败的更大图景，研究结果将比目前的触摸精度模型（只产生目标尺寸和间距指南）具有更广泛的意义。该项目在普遍但尚未探索的界面问题的认知建模方面迈出了一大步。

Touch interfaces on mobile phones and tablets are notoriously error prone. One plausible reason for slow progress in improving the usability of touchscreen devices is that research and design efforts in human-computer interaction take a relatively narrow focus on identifying, understanding, and eliminating human error, focusing primarily on the specific individual touch interactions rather than the broader task knowledge that a person needs to use a device to do a real-world task. This project takes a different perspective, suggesting that usage errors represent breakdowns in a person's efforts to adapt to the complexity of the tasks and devices, and that many of these problems can be traced to interface designers not inadequately considering the cognitive task knowledge that a user will need to do a task. Many touchscreen design guidelines are arbitrary and based on common practice and established user expectations. This project aims to change the shape of research and practice for touch interaction on mobile devices by introducing sound scientific principles of cognitive task analysis and error analysis to the design and analysis of such interfaces. This project could benefit society by dramatically improving the user interface design of touchscreen interfaces.The project will develop cognitive models, which are computer programs that behave in some way like humans. The project will specifically develop cognitive models of touch interaction that could be used by designers when designing new touchscreen interfaces, to identify where users are likely to encounter problems with the new interface. The project includes three stages. The first stage will develop a small set of instrumented interfaces on mobile platforms to collect human data for error-prone tasks. The experiments will examine the variation of the probability of error at specific points during a sequence, the visual feedback provided for specific actions, and the user knowledge of error recovery and avoidance actions. The second stage will follow other successful efforts in cognitive modeling to analyze user data from our exploratory experiment with the goal of developing a representation of the strategies we have observed. An automated process will explore a search space of plausible cognitive strategies to identify models that explain the performance data collected during the first stage. The third stage will validate the hypotheses pertaining to the cognitive strategies that contribute to the successful and unsuccessful use of touchscreen devices that are established during the first two stages. The results will have much wider implications than current models of touch accuracy (which produce little more than target size and spacing guidelines) by focusing on the bigger picture of expertise and failure. The project takes a large step into the cognitive modeling of pervasive but not-yet-unexplored interface problems.