The central insight behind the research is a concept the authors call Critical Decision Points, or CDPs. These are specific states in a shared task where two or more competing human strategies would prescribe clearly different next actions. Imagine a kitchen robot collaborating with a person on a recipe: if two different dishes the person might be preparing require identical actions up to a certain moment, watching the person gives the robot no new information. But if the robot guides the interaction so that the person reaches a junction where making soup demands reaching for a pot while making a salad demands reaching for a cutting board, the very next action unambiguously discloses the goal. CDPs are precisely those junctions, and they maximally reveal intent because the divergent strategies force a distinguishing choice.

To turn this intuition into a computable method, the researchers formalized CDPs using a goal-conditioned policy divergence measure. In plain terms, the team evaluated, for any given state of the world, how differently an optimal policy would behave depending on which goal the human is pursuing. High divergence means the state is highly informative: small differences in behavior at that state separate the candidate goals cleanly. Low divergence means the state tells the robot almost nothing, because all plausible goals call for the same action. By quantifying this divergence, the robot can rank future states by how much information they would generate about the human’s hidden objective.

Measuring informativeness, however, is only half the problem. The robot also has to accomplish the actual task, whether that means assembling ingredients in a kitchen or chasing down an opponent in a game. The team therefore embedded the CDP framework inside a Receding-Horizon Planner, a control architecture rooted in model predictive control that repeatedly looks ahead over a short window of future actions, optimizes a cost function, executes the first action, and then replans. The cost function balances two competing objectives: task progress, meaning the robot keeps moving the joint activity toward completion or toward winning, and information gain, meaning the robot steers toward states that maximize policy divergence and therefore reveal the human’s goal. The receding-horizon structure keeps the planning computationally tractable, because the robot never has to search the full branching tree of future interactions; it only reasons a few steps ahead at a time while still accounting for how its own actions will shape what the human does next.

What makes the work especially notable is that the same unified framework operates in two dramatically different social regimes. In the collaborative setting, the robot and human share a fully observable cooking task, and both parties benefit from smooth coordination. Here, steering the human toward CDPs serves partnership quality: the earlier the robot knows whether its partner is preparing one dish or another, the earlier it can hand over the right utensil, clear the correct counter space, or divide labor sensibly. Faster and more accurate goal inference translates directly into more fluent teamwork, reduced awkward pauses, and fewer mistaken anticipations of what the human wants next.

In the competitive setting, the tables turn entirely. The two agents play a partially observable hide-and-seek game in which the human’s goal is hidden from the robot, and revealing it works to the robot’s advantage rather than the human’s. Even here, the CDP-driven planner proved effective: by maneuvering the human into states where the seeker’s and hider’s strategies prescribe different moves, the robot extracted decisive evidence about the opponent’s hidden objective. The authors frame this in the tradition of opponent modeling, in which an agent builds and refines a predictive model of a rival’s intentions from observed behavior. Provoking opponents into informative situations has precedent in the game-playing literature, but this study is among the first to unify adversarial information gathering and cooperative assistance under a single planning objective.

Evaluation was deliberately thorough. The researchers tested their method in both domains in simulation and then transferred the approach to real robots interacting with actual human participants, a transition that notoriously breaks many planning algorithms due to perception noise, motion constraints, and unpredictable human behavior. Across both simulated and physical trials, the CDP-based planner inferred human goals more accurately and, crucially, earlier than baseline strategies that either pursued the task greedily or gathered information without shaping the human. The earlier a robot can commit to a confident estimate of a partner’s goal, the more useful actions it can take on that partner’s behalf, which is why the timing improvement matters as much as the accuracy improvement.

The implications stretch well beyond kitchens and games. Assistive robots in homes, factories, and hospitals constantly face the problem of deciphering what a person is trying to do, often from ambiguous and fragmentary evidence. A wheelchair-mounted manipulator that can gently reposition a shared workspace so that a user’s next movement disambiguates their intent could provide safer and more responsive support. Warehouse cobots could coordinate with human pickers more fluidly. Even autonomous vehicles, which increasingly reason about the intentions of human drivers, could benefit from planners that recognize which traffic situations are most informative about another driver’s route. The study builds on a growing body of research showing that influence is a two-way street: humans adapt to robots, and robots that account for this adaptation, rather than treating human behavior as fixed background noise, plan fundamentally better.

At the same time, the work raises questions the authors themselves situate within a broader research conversation about robots that influence humans. Steering a person’s choices, even mildly and for legitimate informative purposes, touches on autonomy and transparency, and the competitive results in particular show that the same mathematics that helps a helpful kitchen assistant can empower an adversarial agent. The researchers report no competing interests and ground their contribution firmly in the engineering goal of faster mutual understanding. As robots move from controlled labs into shared human spaces, frameworks like CDPs suggest a future in which machine teammates do not merely watch and guess, but intelligently arrange the world so that human intentions become legible — a capability that promises smoother collaboration, sharper competition, and a clearer science of how people and machines can read each other’s minds in real time.

The study also connects to a longer intellectual lineage in machine learning and control. The idea of learning a reward function or objective from observed behavior traces back to inverse reinforcement learning, which showed that an observer can recover what an agent is optimizing by watching its actions. The present work inverts that logic: instead of passively accumulating observations, the robot actively engineers the observation conditions, choosing its own actions so that the human’s behavior becomes maximally diagnostic. This places the method in the emerging field of influence-aware planning, where an agent treats its partner’s future decisions as part of the environment it can shape rather than as an exogenous process it must simply predict.

Methodologically, the reliance on a receding-horizon architecture reflects decades of practice in model predictive control, a technique originally developed in industrial process control and now widespread in robotics because it handles constraints and replans gracefully under uncertainty. Adapting this mature machinery to a social objective — trading off task progress against epistemic value — is a notable engineering contribution, since classical cost functions rarely encode anything like information gain about a partner’s intent.

The dual-domain evaluation also strengthens the claim of generality. Many goal-recognition systems are validated in a single task, leaving open whether the underlying measure of informativeness is task-specific. By demonstrating that the same policy-divergence criterion improves inference in a fully observable cooperative setting and a partially observable adversarial one, the authors provide evidence that Critical Decision Points are a domain-independent structural property of sequential decision problems, not an artifact of a particular scenario. The consistency of results across simulation and physical hardware further suggests the approach is robust to the perception noise and behavioral variability that typically degrade deployed interactive planners.

Subject of Research: Robot planning strategies that influence human behavior to accelerate goal inference during collaboration and competition.

Article Title: Robots influencing humans to reveal their goals during collaboration and competition

Article References: Ghose, D., Gitelson, O., Lewkowicz, M., Brawer, J., Roncone, A., Vazquez, M., & Scassellati, B. (2026). Robots influencing humans to reveal their goals during collaboration and competition. Autonomous Robots, 50(3), Article 38. https://doi.org/10.1007/s10514-026-10267-2

Image Credits: AI Generated

DOI: 10.1007/s10514-026-10267-2

Keywords: human-robot interaction, goal inference, critical decision points, receding horizon planning, social robotics, opponent modeling, human-robot collaboration, game theory, artificial intelligence, planning, shared autonomy, Robots

Tetris may look simple, but successful play requires the brain to perform several operations at once. Players must monitor a falling shape, rotate it mentally, predict where it will fit, remember the current board configuration, track upcoming pieces, and rapidly transform those decisions into finger movements. As the game becomes faster, these processes must operate under intense time pressure. This combination makes Tetris very different from traditional laboratory tasks that isolate a single ability, such as the Stroop task for cognitive control or the n-back task for working memory. Tetrix preserves the game’s integrated demands while allowing researchers to separate its visual, motor, and cognitive components experimentally.

The project was developed by modifying an open-source Python implementation of Tetris using the Pygame library and then integrating the game into PsychoPy, a widely used platform for behavioral and neuroimaging experiments. Its architecture separates the game mechanics from the broader experimental design, allowing investigators to alter settings through configuration files rather than rewriting the entire program. Researchers can define the starting level, control the rate at which blocks fall, determine how many completed lines are needed to advance, adjust scoring rules, and prevent level progression when a constant difficulty is required. They can also select whether the next one, two, or three blocks appear on screen, or remove the preview entirely to reduce visuospatial planning during control conditions.

The program also includes several components that can run in parallel through Python’s multiprocessing framework. A pretrial version measures individual performance and can be used as a standalone behavioral task. A main gameplay process is intended for neuroimaging experiments, while a visually simplified “watching” process displays falling blocks without allowing participants to control them. The paradigm records scanner trigger signals, keypresses, timing information, game events, and performance variables in log files. Researchers can pseudorandomize block sequences and experimental conditions using fixed random seeds, ensuring that the same stimuli can be reproduced across participants or testing sessions. This reproducibility is particularly important in fMRI, where small differences in timing or stimulus content can affect the measured blood-oxygen-level-dependent signal.

Tetrix is built around a set of control conditions designed to identify which parts of Tetris gameplay drive brain activity. In the default design, participants first complete practice rounds so that the game can estimate an appropriate difficulty level. During the main experiment, they play Tetris for 30 seconds, followed by one of three conditions: watching an automated version of the game, making button presses without playing, or viewing a fixation cross as a baseline. The visual control presents blocks that move independently of the participant’s actions and do not stack, while the motor control displays symbols indicating when participants should alternate button presses. Comparing gameplay with these conditions helps researchers distinguish activity related to complex visuospatial operations from activity caused simply by seeing moving shapes, pressing buttons, or maintaining a resting baseline.

To demonstrate that the system could work inside an MRI scanner, the researchers conducted a pilot study involving seven participants. One participant was excluded because strong head motion caused a field-of-view shift, leaving six datasets for the main neuroimaging analysis. Participants completed 21 gameplay trials, each lasting 30 seconds, with variable intervals of six to eight seconds between blocks. Scanning was performed on a 3-Tesla MRI system using a multiband echo-planar imaging sequence with a repetition time of 1.2 seconds. The functional images covered the whole brain at a resolution of approximately 3 millimeters in-plane and 3.3 millimeters through-plane, while a high-resolution T1-weighted anatomical scan was collected for each participant.

The researchers processed the data with SPM12, a standard software package for statistical parametric mapping. Their preprocessing pipeline included motion estimation, correction of outlier volumes, slice-timing correction, anatomical-functional co-registration, tissue segmentation, normalization to the MNI template, and spatial smoothing with an 8-millimeter Gaussian kernel. Motion parameters were included in the statistical model, and an interpolation procedure called SPIKECOR was used to replace unusually affected volumes. The critical analysis tested whether gameplay produced greater activity than watching Tetris, button pressing, and baseline fixation simultaneously. This conjunction contrast was intended to isolate neural responses associated with the distinctive cognitive demands of playing rather than with basic vision or hand movements.

The resulting activation pattern centered on a distributed frontoparietal network. Bilateral regions in the middle and superior frontal gyri, including areas associated with the frontal eye fields, became active during gameplay. Strong responses also appeared in the posterior parietal cortex, including the superior parietal lobule and intraparietal sulcus, as well as the left middle occipital cortex and parts of the right cerebellum. The frontal eye fields and posterior parietal cortex are major components of the dorsal attention network, which helps direct attention toward relevant locations and coordinate goal-driven visual exploration. In Tetris, these regions may support the rapid selection of important board elements, the monitoring of falling pieces, and the shifting of attention between the current block, the playfield, and the preview window.

The authors argue that the same frontoparietal system may also support visuospatial working memory and mental imagery. Players must retain the shape and orientation of Tetrominoes, imagine possible rotations, and compare those imagined configurations with available spaces on the board. The occipital activation that remained after comparison with the visual control condition may reflect top-down modulation of visual processing, although the researchers caution that eye movements could also contribute. Without eye tracking, it is impossible to determine whether the frontal eye-field response reflects cognitive control, differences in saccade frequency, or both. Cerebellar activity may likewise reflect more than simple finger movement, potentially involving movement coordination and predictions about the sensory consequences of rapid actions.

The study also reports voxel-wise Hedges’ g effect-size maps that may help future laboratories estimate sample sizes, although the authors emphasize that the pilot sample is too small for definitive conclusions. Some estimated effects were exceptionally large, exceeding g = 5, a result that can occur when a small sample produces strong but unstable group statistics. An additional group of ten participants showed broadly similar activation clusters, offering preliminary replication, but the study was not designed to establish precise causal roles for the identified regions. Head-motion spikes occurred across participants, underscoring a major challenge for MRI research using physically demanding games. Even with correction and interpolation, frequent hand movements may produce subtle body and head displacement that can contaminate neural measurements.

Tetrix is also connected to a growing clinical interest in Tetris-based interventions. Previous studies have suggested that playing a visuospatial game after trauma may reduce later intrusive memories, possibly by competing with the mental imagery and visuospatial working-memory resources involved in forming or reconsolidating traumatic memories. Tetris-based interventions have been examined in emergency departments, experimental trauma studies, and clinical populations with post-traumatic stress disorder. The new pilot findings raise the possibility that the game’s effects depend not only on working-memory load but also on rapid visuospatial reorientation and sustained engagement of the dorsal attention network. That interpretation remains hypothetical, however, and the present study did not test treatment outcomes or patients with PTSD.

The authors describe Tetrix as an ongoing project rather than a finished clinical instrument. Later versions added adjustable trial lengths, optional experimental blocks, detailed gameplay recording, motor-condition logging, and replay-based controls that can reproduce the timing of earlier gameplay. Such features could allow future studies to manipulate one variable at a time, including game speed, preview-window size, level progression, or motor demands. These experiments may clarify whether Tetris-related brain activity reflects working-memory capacity, mental rotation, attention shifting, motor planning, reward processing, or the interaction of all these functions. For now, Tetrix offers researchers an unusually accessible bridge between a popular game and rigorous cognitive neuroscience: a reproducible, configurable task that can be downloaded, modified, and tested across laboratories and clinical settings.

Subject of Research: A standardized Tetris-based behavioral and fMRI paradigm for studying attention, visuospatial working memory, mental imagery, motor planning, and related neural networks.

Article Title: Tetrix: A novel Tetris-based paradigm for neuroimaging research and clinical applications

Article References: Grote, J., Stocker, J. E., Sommer, J., Hamm, A.-M., Kessler, H., & Jansen, A. (2026). Tetrix: A novel Tetris-based paradigm for neuroimaging research and clinical applications. Behavior Research Methods, 58, Article 279. https://doi.org/10.3758/s13428-026-03150-6

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03150-6

Keywords: Tetris, Tetrix, fMRI, PsychoPy, visuospatial working memory, mental imagery, dorsal attention network, cognitive control, motor planning, neuroimaging, PTSD research