Affective computing research has advanced emotion recognition systems using facial expressions, voices, gaits, and physiological signals, yet these methods are often impractical. This study integrates mouse cursor motion analysis into affective computing and investigates the idea that movements of the computer cursor can provide information about emotion of the computer user. We extracted 16–26 trajectory features during a choice‐reaching task and examined the link between emotion and cursor motions. Participants were induced for positive or negative emotions by music, film clips, or emotional pictures, and they indicated their emotions with questionnaires. Our 10‐fold cross‐validation analysis shows that statistical models formed from “known” participants (training data) couldpredictnearly 10%–20% of the variance of positive affect and attentiveness ratings of “unknown” participants, suggesting that cursor movement patterns such as the area under curve and direction change help infer emotions of computer users.

How do people use category membership and similarity for making inductive inferences? The authors addressed this question by examining the impact of category labels and category features on inference and classification tasks that were designed to be comparable. In the inference task, participants predicted the value of a missing feature of an item given its category label and other feature values. In the classification task, participants predicted the category label of an item given its feature values. The results from 4 experiments suggest that category membership influences inference even when similarity information contradicts the category label. This tendency was stronger when the category label conveyed class inclusion information than when the label reflected a feature of the category. These findings suggest that category membership affects inference beyond similarity and that category labels and category features are 2 different things.

Recent evidence suggests that top-down attention facilitates unconscious semantic processing. To clarify the role of attention in unconscious semantic processing, we traced trajectories of the computer mouse in a semantic priming task and scrutinized the extent to which top-down attention enhances unconscious semantic processing in four different stimulus-onset asynchrony (SOA: 50, 200, 500, or 1000ms) conditions. Participants judged whether a target digit (e.g., “6”) was larger or smaller than five, preceded by a masked priming digit (e.g., “9”). The pre-prime duration changed randomly from trial to trial to disrupt participants’ top-down attention in an uncued condition (in a cued condition, a green square cue was presented to facilitate participants’ top-down attention). The results show that top-down attention modifies the time course of subliminal semantic processing, and the temporal attention window lasts more than 1000ms; attention facilitated by the cue may amplify semantic priming to some extent, yet the amplification effect of attention is relatively minor.

The accurate detection of attention-deficit/hyperactivity disorder (ADHD) symptoms, such as inattentiveness and behavioral disinhibition, is crucial for delivering timely assistance and treatment. ADHD is commonly diagnosed and studied with specialized questionnaires and behavioral tests such as the stop-signal task. However, in cases of late-onset or mild forms of ADHD, behavioral measures often fail to gauge the deficiencies well-highlighted by questionnaires. To improve the sensitivity of behavioral tests, we propose a novel version of the stop-signal task (SST), which integrates mouse cursor tracking. In two studies, we investigated whether introducing mouse movement measures to the stop-signal task improves associations with questionnaire-based measures, as compared to the traditional (keypress-based) version of SST. We also scrutinized the influence of different parameters of stop-signal tasks, such as the method of stop-signal delay setting or definition of response inhibition failure, on these associations. Our results show that a) SSRT has weak association with impulsivity, while mouse movement measures have strong and significant association with impulsivity; b) machine learning models trained on the mouse movement data from “known” participants using nested cross-validation procedure can accurately predict impulsivity ratings of “unknown” participants; c) mouse movement features such as maximum acceleration and maximum velocity are among the most important predictors for impulsivity; d) using preset stop-signal delays prompts behavior that is more indicative of impulsivity.