Robustness of performance during domain change in an esport: A study of within-expertise transfer.

Research on the transfer of skill from the circumstances in which it was learned to partially or completely novel tasks or situations is a foundational topic in the study of learning, memory, education, and expertise. A long history of transfer research has led to the conclusion that skill learning is generally domain specific. One important transfer problem occurs when a domain of expertise undergoes a fundamental shift, as when experts must adapt to changes in technology, rules, or professional practice. Here we examine skill maintenance in StarCraft 2, a video game of skills which undergoes frequent changes due to updates and includes a variety of gameplay options. Of particular interest are two competing predictions about how transfer will interact with expertise in this domain. The first approach emphasizes perceived similarity of the domains and predicts that skilled individuals will exhibit more favourable transfer than novices as these people will know enough to avoid processes, methods, and strategies which no longer apply after a domain change. The second emphasizes maximal adaptation to task constraints and predicts that experts will suffer the most during a domain change because of the loss of exploitable affordances. Neither approach did a good job explaining behaviour after the major game update called 'StarCraft 2: Heart of the Swarm,' perhaps because transfer was generally strong across all players. However, when examining transfer in the context of larger changes to gameplay, transfer seemed slightly better in more experienced players. The theoretical implications of this apparent interaction effect, and of the apparent resilience of more experienced StarCraft 2 players to transfer costs, are discussed.

Grapheme-color synaesthesia benefits rule-based Category learning.

Researchers have long suspected that grapheme-color synaesthesia is useful, but research on its utility has so far focused primarily on episodic memory and perceptual discrimination. Here we ask whether it can be harnessed during rule-based Category learning. Participants learned through trial and error to classify grapheme pairs that were organized into categories on the basis of their associated synaesthetic colors. The performance of synaesthetes was similar to non-synaesthetes viewing graphemes that were physically colored in the same way. Specifically, synaesthetes learned to categorize stimuli effectively, they were able to transfer this learning to novel stimuli, and they falsely recognized grapheme-pair foils, all like non-synaesthetes viewing colored graphemes. These findings demonstrate that synaesthesia can be exploited when learning the kind of material taught in many classroom settings.

LAG-1: A dynamic, integrative model of learning, attention, and gaze.

It is clear that learning and attention interact, but it is an ongoing challenge to integrate their psychological and neurophysiological descriptions. Here we introduce LAG-1, a dynamic neural field model of learning, attention and gaze, that we fit to human learning and eye-movement data from two category learning experiments. LAG-1 comprises three control systems: one for visuospatial attention, one for saccadic timing and control, and one for category learning. The model is able to extract a kind of information gain from pairwise differences in simple associations between visual features and categories. Providing this gain as a reentrant signal with bottom-up visual information, and in top-down spatial priority, appropriately influences the initiation of saccades. LAG-1 provides a moment-by-moment simulation of the interactions of learning and gaze, and thus simultaneously produces phenomena on many timescales, from the duration of saccades and gaze fixations, to the response times for trials, to the slow optimization of attention toward task relevant information across a whole experiment. With only three free parameters (learning rate, trial impatience, and fixation impatience) LAG-1 produces qualitatively correct fits for learning, behavioural timing and eye movement measures, and also for previously unmodelled empirical phenomena (e.g., fixation orders showing stimulus-specific attention, and decreasing fixation counts during feedback). Because LAG-1 is built to capture attention and gaze generally, we demonstrate how it can be applied to other phenomena of visual cognition such as the free viewing of visual stimuli, visual search, and covert attention.

Informed inferences of unknown feature values in categorization.

Many current computational models of object categorization either include no explicit provisions for dealing with incomplete stimulus information (e.g. Kruschke, Psychological Review 99:22-44, 1992) or take approaches that are at odds with evidence from other fields (e.g. Verguts, Ameel, & Storms, Memory & Cognition 32:379-389, 2004). In two experiments centered around the inverse base-rate effect, we demonstrate that people not only make highly informed inferences about the values of unknown features, but also subsequently use the inferred values to come to a categorization decision. The inferences appear to be based on immediately available information about the particular stimulus under consideration, as well as on higher-level inferences about the stimulus class as a whole. Implications for future modeling efforts are discussed.

The ubiquity of selective attention in the processing of feedback during category learning.

Feedback is essential for many kinds of learning, but the cognitive processes involved in learning from feedback are unclear. Models of category learning incorporate selective attention to stimulus features while generating a response, but during the feedback phase of an experiment, it is assumed that participants receive complete information about stimulus features as well as the correct category. The present work looks at eye tracking data from six category learning datasets covering a variety of category complexities and types. We find that selective attention to task-relevant information is pervasive throughout feedback processing, suggesting a role for selective attention in memory encoding of category exemplars. We also find that error trials elicit additional stimulus processing during the feedback phase. Finally, our data reveal that participants increasingly skip the processing of feedback altogether. At the broadest level, these three findings reveal that selective attention is ubiquitous throughout the entire category learning task, functioning to emphasize the importance of certain stimulus features, the helpfulness of extra stimulus encoding during times of uncertainty, and the superfluousness of feedback once one has learned the task. We discuss the implications of our findings for modelling efforts in category learning from the perspective of researchers trying to capture the full dynamic interaction of selective attention and learning, as well as for researchers focused on other issues, such as category representation, whose work only requires simplifications that do a reasonable job of capturing learning.

Extremely selective attention: eye-tracking studies of the dynamic allocation of attention to stimulus features in categorization.

Humans have an extremely flexible ability to categorize regularities in their environment, in part because of attentional systems that allow them to focus on important perceptual information. In formal theories of categorization, attention is typically modeled with weights that selectively bias the processing of stimulus features. These theories make differing predictions about the degree of flexibility with which attention can be deployed in response to stimulus properties. Results from 2 eye-tracking studies show that humans can rapidly learn to differently allocate attention to members of different categories. These results provide the first unequivocal demonstration of stimulus-responsive attention in a categorization task. Furthermore, the authors found clear temporal patterns in the shifting of attention within trials that follow from the informativeness of particular stimulus features. These data provide new insights into the attention processes involved in categorization.

Classic motor chunking theory fails to account for behavioural diversity and speed in a complex naturalistic task.

In tasks that demand rapid performance, actions must be executed as efficiently as possible. Theories of expert motor performance such as the motor chunking framework suggest that efficiency is supported by automatization, where many serial actions are automatized into smaller chunks, or groups of commonly co-occuring actions. We use the fast-paced, professional eSport StarCraft 2 as a test case of the explanatory power of the motor chunking framework and assess the importance of chunks in explaining expert performance. To do so, we test three predictions motivated by a simple motor chunking framework. (1) StarCraft 2 players should exhibit an increasing number of chunks with expertise. (2) The proportion of actions falling within a chunk should increase with skill. (3) Chunks should be faster than non-chunks containing the same atomic behaviours. Although our findings support the existence of chunks, they also highlight two problems for existing accounts of rapid motor execution and expert performance. First, while better players do use more chunks, the proportion of actions within a chunks is stable across expertise and expert sequences are generally more varied (the diversity problem). Secondly, chunks, which are supposed to enjoy the most extreme automatization, appear to save little or no time overall (the time savings problem). Instead, the most parsimonious description of our latency analysis is that players become faster overall regardless of chunking.

Using Video Game Telemetry Data to Research Motor Chunking, Action Latencies, and Complex Cognitive-Motor Skill Learning.

Many theories of complex cognitive-motor skill learning are built on the notion that basic cognitive processes group actions into easy-to-perform sequences. The present work examines predictions derived from laboratory-based studies of motor chunking and motor preparation using data collected from the real-time strategy video game StarCraft 2. We examined 996,163 action sequences in the telemetry data of 3,317 players across seven levels of skill. As predicted, the latency to the first action (thought to be the beginning of a chunked sequence) is delayed relative to the other actions in the group. Other predictions, inspired by the memory drum theory of Henry and Rogers, received only weak support.

Over the hill at 24: persistent age-related cognitive-motor decline in reaction times in an ecologically valid video game task begins in early adulthood.

Typically studies of the effects of aging on cognitive-motor performance emphasize changes in elderly populations. Although some research is directly concerned with when age-related decline actually begins, studies are often based on relatively simple reaction time tasks, making it impossible to gauge the impact of experience in compensating for this decline in a real world task. The present study investigates age-related changes in cognitive motor performance through adolescence and adulthood in a complex real world task, the real-time strategy video game StarCraft 2. In this paper we analyze the influence of age on performance using a dataset of 3,305 players, aged 16-44, collected by Thompson, Blair, Chen & Henrey [1]. Using a piecewise regression analysis, we find that age-related slowing of within-game, self-initiated response times begins at 24 years of age. We find no evidence for the common belief expertise should attenuate domain-specific cognitive decline. Domain-specific response time declines appear to persist regardless of skill level. A second analysis of dual-task performance finds no evidence of a corresponding age-related decline. Finally, an exploratory analyses of other age-related differences suggests that older participants may have been compensating for a loss in response speed through the use of game mechanics that reduce cognitive load.

Learning-induced changes in attentional allocation during categorization: a sizable catalog of attention change as measured by eye movements.

Learning how to allocate attention properly is essential for success at many categorization tasks. Advances in our understanding of learned attention are stymied by a chicken-and-egg problem: there are no theoretical accounts of learned attention that predict patterns of eye movements, making data collection difficult to justify, and there are not enough datasets to support the development of a rich theory of learned attention. The present work addresses this by reporting five measures relating to the overt allocation of attention across 10 category learning experiments: accuracy, probability of fixating irrelevant information, number of fixations to category features, the amount of change in the allocation of attention (using a new measure called Time Proportion Shift - TIPS), and a measure of the relationship between attention change and erroneous responses. Using these measures, the data suggest that eye-movements are not substantially connected to error in most cases and that aggregate trial-by-trial attention change is generally stable across a number of changing task variables. The data presented here provide a target for computational models that aim to account for changes in overt attentional behaviors across learning.

Waiting and weighting: Information sampling is a balance between efficiency and error-reduction.

The current study investigates the relative extent to which information utility and planning efficiency guide information-sampling strategies in a classification task. Prior research has pointed to the importance of probability gain, the degree to which sampling a feature reduces the chance of error, in contexts where participants are restricted to one sample. We monitored participants as they sampled information in an unrestricted context and recorded whether they began their search with a high gain feature or an efficient feature that ultimately allowed for fewer samples per trial. Participants preferred to sample the more efficient feature first, especially when feature information had a higher access cost (Experiment 1). When access costs were all but eliminated using eye-tracking (Experiment 2), participants' fixations still emphasized efficiency over high probability gain, though probability gain was shown to influence access patterns.

Temporal characteristics of overt attentional behavior during category learning.

Many theories of category learning incorporate mechanisms for selective attention, typically implemented as attention weights that change on a trial-by-trial basis. This is because there is relatively little data on within-trial changes in attention. We used eye tracking and mouse tracking as fine-grained measures of attention in three complex visual categorization tasks to investigate temporal patterns in overt attentional behavior within individual categorization decisions. In Experiments 1 and 2, we recorded participants' eye movements while they performed three different categorization tasks. We extended previous research by demonstrating that not only are participants less likely to fixate irrelevant features, but also, when they do, these fixations are shorter than fixations to relevant features. We also found that participants' fixation patterns show increasingly consistent temporal patterns. Participants were faster, although no more accurate, when their fixation sequences followed a consistent temporal structure. In Experiment 3, we replicated these findings in a task where participants used mouse movements to uncover features. Overall, we showed that there are important temporal regularities in information sampling during category learning that cannot be accounted for by existing models. These can be used to supplement extant models for richer predictions of how information is attended to during the buildup to a categorization decision.

Video game telemetry as a critical tool in the study of complex skill learning.

Cognitive science has long shown interest in expertise, in part because prediction and control of expert development would have immense practical value. Most studies in this area investigate expertise by comparing experts with novices. The reliance on contrastive samples in studies of human expertise only yields deep insight into development where differences are important throughout skill acquisition. This reliance may be pernicious where the predictive importance of variables is not constant across levels of expertise. Before the development of sophisticated machine learning tools for data mining larger samples, and indeed, before such samples were available, it was difficult to test the implicit assumption of static variable importance in expertise development. To investigate if this reliance may have imposed critical restrictions on the understanding of complex skill development, we adopted an alternative method, the online acquisition of telemetry data from a common daily activity for many: video gaming. Using measures of cognitive-motor, attentional, and perceptual processing extracted from game data from 3360 Real-Time Strategy players at 7 different levels of expertise, we identified 12 variables relevant to expertise. We show that the static variable importance assumption is false--the predictive importance of these variables shifted as the levels of expertise increased--and, at least in our dataset, that a contrastive approach would have been misleading. The finding that variable importance is not static across levels of expertise suggests that large, diverse datasets of sustained cognitive-motor performance are crucial for an understanding of expertise in real-world contexts. We also identify plausible cognitive markers of expertise.

Errors, efficiency, and the interplay between attention and category learning.

Learning to identify objects as members of categories is an essential cognitive skill and learning to deploy attention effectively is a core component of that process. The present study investigated an assumption imbedded in formal models of categorization: error is necessary for attentional learning. Eye-trackers were used to record participants' allocation of attention to task relevant and irrelevant features while learning a complex categorization task. It was found that participants optimized their fixation patterns in the absence of both performance errors and corrective external feedback. Optimization began immediately after each category was mastered and continued for many trials. These results demonstrate that error is neither necessary nor sufficient for all forms of attentional learning.