How does expertise influence the perception of representational and abstract paintings? We asked 20 experts on art history and 20 laypersons to explore and evaluate a series of paintings ranging in style from representational to abstract in five categories. We compared subjective esthetic judgments and emotional evaluations, gaze patterns, and electrodermal reactivity between the two groups of participants. The level of abstraction affected esthetic judgments and emotional valence ratings of the laypersons but had no effect on the opinions of the experts: the laypersons’ esthetic and emotional ratings were highest for representational paintings and lowest for abstract paintings, whereas the opinions of the experts were independent of the abstraction level. The gaze patterns of both groups changed as the level of abstraction increased: the number of fixations and the length of the scanpaths increased while the duration of the fixations decreased. The viewing strategies – reflected in the target, location, and path of the fixations – however indicated that experts and laypersons paid attention to different aspects of the paintings. The electrodermal reactivity did not vary according to the level of abstraction in either group but expertise was reflected in weaker responses, compared with laypersons, to information received about the paintings.

As AI becomes more common in everyday living, there is an increasing demand for intelligent systems that are both performant and understandable. Explainable AI (XAI) systems aim to provide comprehensible explanations of decisions and predictions. At present, however, evaluation processes are rather technical and not sufficiently focused on the needs of human users. Consequently, evaluation studies involving human users can serve as a valuable guide for conducting user studies. This paper presents a comprehensive review of 65 user studies evaluating XAI systems across different domains and application contexts. As a guideline for XAI developers, we provide a holistic overview of the properties of XAI systems and evaluation metrics focused on human users (human-centered). We propose objectives for the human-centered design (design goals) of XAI systems. To incorporate users' specific characteristics, design goals are adapted to users with different levels of AI expertise (AI novices and data experts). In this regard, we provide an extension to existing XAI evaluation and design frameworks. The first part of our results includes the analysis of XAI system characteristics. An important finding is the distinction between the core system and the XAI explanation, which together form the whole system. Further results include the distinction of evaluation metrics into affection towards the system, cognition, usability, interpretability, and explanation metrics. Furthermore, the users, along with their specific characteristics and behavior, can be assessed. For AI novices, the relevant extended design goals include responsible use, acceptance, and usability. For data experts, the focus is performance-oriented and includes human-AI collaboration and system and user task performance.

The knowledge about the usual size of objects—familiar size—is known to be a taken into account for distance perception. The influence of familiar size on action programming is less clear and has not yet been tested with regard to vergence eye movements. In two experiments, we stereoscopically presented everyday objects, such as a credit card or a package of paper tissues, and varied the distance as specified by binocular disparity and the distance as specified by familiar size. Participants had to fixate the shown object and subsequently reach towards it either with open or with closed eyes. When binocular disparity and familiar size were in conflict, reaching movements revealed a combination of the two depth cues with individually different weights. The influence of familiar size was larger when no visual feedback was available during the reaching movement. Vergence movements closely followed binocular disparity and were largely unaffected by familiar size. In sum, the results suggest that in this experimental setting familiar size is taken into account for programming and executing reaching movements while vergence movements are primarily based on binocular disparity.