ProBackend
Psychology & Behaviour
4 days ago6 min read

AI Impact on Human Psychology: What Counter-Strike Experts’ 90ms Edge Reveals

AI Impact on Human Psychology: What Counter-Strike Experts’ 90ms Edge Reveals

A study of more than 300 Counter-Strike players uses drift-diffusion modelling to separate faster visual encoding, evidence accumulation and motor execution from multitasking ability—and shows what computational models can and cannot tell us about psychology.

AI impact on human psychology: what a 90ms gaming advantage means

Can expertise in a fast, visually demanding game change how quickly people make decisions? A University of Sheffield study of more than 300 Counter-Strike players offers a carefully bounded answer: expert players were about 90 milliseconds faster than novice players on visual decision tasks. Drift-diffusion modelling helped researchers distinguish parts of that advantage, including faster visual encoding, faster accumulation of evidence in difficult choices and quicker motor responses. But experts did not show an advantage when tasks required multitasking.

The result is relevant to the AI impact on human psychology because it illustrates how computational models can help researchers describe decision processes—not because the study used AI to read minds or establish that gaming causes lasting psychological change. A model estimates latent stages from patterns of responses. Its usefulness depends on the task, assumptions and evidence behind it.

What the study measured

The research, reported by Neuroscience News and published in Experimental Psychology: Human Perception and Performance, compared players across a range of experience, from novices to experts with more than 5,000 hours of gameplay. In standardized visual reaction tasks, the expert group responded on average around 90ms faster than novices.

That average is a group-level difference, not a guarantee that every expert beats every novice, nor a universal measure of intelligence. A fraction of a second can matter in time-critical settings, but translating a laboratory task result into safer driving or better emergency performance would require separate evidence. The study itself does not show that gamers are better drivers, emergency responders or decision-makers in every context.

The distinction between a fast response and a good decision also matters. Reaction time combines several processes: noticing and encoding a stimulus, deciding what it means, selecting a response and physically carrying it out. A shorter time by itself does not identify which part changed. It may also involve a speed–accuracy trade-off, so accuracy and response speed must be considered together.

What is AI in psychology? A model of decisions, not a mind reader

In psychology, AI and computational methods can be used to analyze behavior, test predictions and compare formal explanations of how people make choices. Drift-diffusion modelling (DDM) is a mathematical framework used for choice and response-time data. It represents a decision as evidence building over time toward one of two response boundaries. The model can estimate the rate at which evidence accumulates, the amount of evidence required before responding, and time spent in processes outside the central decision, such as sensory encoding and motor execution.

DDM is not synonymous with modern generative AI, and the presence of a mathematical model does not mean a system directly observes thoughts. In this study, modelling provides a way to decompose observed performance into theoretically meaningful components. These components are estimates, not direct measurements of individual neurons or a complete account of a person’s psychology.

This is one useful example of artificial intelligence and human psychology intersecting: computational approaches can make competing explanations more explicit and testable. A model may indicate that a group’s faster responses are consistent with more rapid evidence accumulation, but that conclusion remains conditional on the model and experimental design. It does not determine why those participants developed that profile.

Where the advantage appeared—and where it did not

The reported expert advantage was distributed across several stages. Experts encoded visual information faster, accumulated evidence more rapidly when making complex or ambiguous choices, and executed motor responses more quickly. That pattern is more informative than the headline reaction-time gap alone: it suggests that the difference is not simply a matter of pressing a key sooner after an otherwise identical decision.

At the same time, the study found no measurable expert advantage on multitasking tasks. This is an important limit on popular claims that action-game experience produces a broad ability to handle many demands at once. The evidence instead points toward faster focused processing under the tested conditions. Efficiently responding to one relevant stream of information is not the same skill as dividing attention among simultaneous streams.

The result also cautions against treating “processing speed” as one general capacity. Performance can depend on what information is presented, how ambiguous it is, what response is required and whether attention is divided. A strength in a visual choice task need not transfer to unrelated forms of reasoning or everyday judgment.

Can AI understand human psychology?

Not in the human sense of understanding a person’s experiences, motives or inner life. A computational model can capture patterns in behavior and help evaluate hypotheses about underlying processes. That is valuable, but it is narrower than comprehending a person. The Counter-Strike findings show how models can help researchers ask where response-time differences arise; they do not reveal a player’s thoughts or explain the personal meaning of gaming.

There are at least two explanations for the association between expertise and faster performance. Practice might sharpen skills relevant to the game, or people who already have faster visual processing might be more likely to become highly engaged players. The reported study cannot establish which explanation is correct. As lead researcher Eleanor Hyde noted, the available data do not establish whether speed results from extensive play or whether people with naturally faster processing gravitate toward gaming.

To determine causation, researchers would need designs that track change over time, such as controlled training studies with suitable comparison groups, and assess whether any gains persist or transfer beyond the practiced tasks. The possibility of training older adults or applying these findings to high-stakes work is therefore a question for future research, not a demonstrated intervention or selection tool.

What the result means for AI and behavioral decision modelling

The article’s relevance to AI-behavioral decision modelling is methodological. Formal models can separate candidate stages of a decision and generate more precise questions than a single overall reaction-time score. They also help expose the limits of a claim: a faster evidence-accumulation estimate does not by itself prove a general cognitive advantage, identify its cause or predict performance in every real-world situation.

A related research thread, [1904.06866v1] Predicting human decisions with behavioral theories, reflects the broader aim of using behavioral theory to make predictions about choices. The general lesson is to treat prediction as an empirical task: compare model predictions with observed behavior, examine errors and test whether findings generalize. Neither a model nor a high predictive score should be mistaken for a complete psychological explanation.

For anyone asking “best AI for human psychology,” there is no single tool that can replace a well-designed study, validated measures and careful interpretation. The appropriate method depends on the question and data. In this case, a drift-diffusion model is useful because the researchers wanted to parse response time into plausible components. It is not a diagnostic assessment of players and should not be used to infer an individual’s character or suitability for a profession.

What researchers are investigating next

The Sheffield team is comparing tactical first-person-shooter players with players of League of Legends, a multiplayer online battle arena game that places different demands on players, including longer-range strategy and team resource decisions. Comparing genres may help test whether cognitive profiles correspond to particular game demands rather than gaming experience in general.

Planned work involving mobile eye-tracking and 64-channel EEG could add information about visual search and brain activity during gameplay. Such measures may help connect response patterns with the timing of attention and neural responses, but they will still require careful interpretation. Additional measurements do not automatically prove causation or demonstrate transfer to everyday tasks.

The balanced conclusion is that experienced Counter-Strike players in this study showed a meaningful average speed advantage on visual decision tasks, with modelled differences spanning sensory, decision and motor stages. They did not show better multitasking. The findings offer a focused case study in how behavioral models can illuminate human performance while leaving open the questions of cause, transfer and individual psychological experience.

Source