Embodiment & Emergence — Weekly Roundup #6
May 1, 2026 · agency detection · pattern completion · behavioral states · cortical dynamics · network emergence
Introduction
This issue brings together work on agency, perception, behavior, and collective organization, with a shared focus on how systems operate under constraint. Across domains that include sensorimotor control, visual inference, spontaneous behavior, large-scale brain dynamics, and social networks, the studies examine how patterns are detected, completed, stabilized, and reorganized. What links them is not a single mechanism, but a recurring condition: systems do not respond to the full set of available inputs, but to what they are able to register, organize, and sustain as a working model of the situation.
Several of the pieces make this constraint visible in different ways. In sensorimotor tasks, sensitivity to losing control is sharper than sensitivity to gaining it. In visual cortex, partial input is actively structured into coherent percepts that may not correspond to what is present. In behavior, what appears spontaneous is organized into longer-lasting, context-sensitive states. In cortical dynamics, local signals are embedded within large-scale patterns that dominate activity across space and frequency. In social systems, local interaction rules accumulate into network structures that feed back into what individuals can perceive and do.
From the perspective of embodiment, this concerns how sensing, movement, and physiological processes constrain what enters into the system’s model of the situation. Information that is not registered or integrated does not factor into that model, even if it is present. From the perspective of emergence, it concerns whether coordination can update as conditions shift, or whether it remains tied to established patterns that continue to reproduce themselves. The material in this issue does not resolve that question, but it makes clearer what such updating would depend on.
Subscribe for a weekly synthesis of research on mind, body, and collective behavior.
Perceptual sensitivity to changes in control depends on direction, but metacognitive monitoring does not
Takada, K., Wen, W., Kasahara, S., & colleagues. (2026). Perceptual sensitivity, but not metacognitive monitoring, is shaped by increases and decreases in control. Experimental Brain Research, 244, 102. doi.org/10.1007/s00221-026-07306-w
TAGS: sense of agency, error monitoring, regularity detection, metacognition, m-ratio, signal detection theory, motor control, online experiment
OVERVIEW: This study examined how participants detect changes in their level of control over a moving object and how accurately they evaluate their own performance. Participants steered a dot on a screen, reported whether their level of control changed during each trial, and then rated their confidence in that judgment. The analysis compared trials in which control decreased with those in which it increased. Across two experiments, participants showed higher perceptual sensitivity to decreases in control than to increases. However, metacognitive efficiency, measured using the m-ratio (a signal detection–based metric of how well confidence tracks actual correctness), did not differ between these conditions.
OF NOTE: The study separates two processes that are often grouped together in discussions of the sense of agency. One is error detection, where the system registers a mismatch between expected and actual outcomes, such as when an action produces less effect than predicted. The other is regularity detection, where the system identifies stable relationships between actions and their consequences, such as noticing that a new pattern of control has become reliable. The results suggest that these processes are not equally sensitive. Loss of control is detected more readily, which is consistent with the idea that detecting breakdowns or mismatches is a more immediate signal for the system. Gains in control require detecting a new pattern over time, which may be inherently slower or less salient. At the same time, the fact that metacognitive efficiency does not differ suggests that the system responsible for evaluating performance operates across both types of input in a relatively uniform way. This supports an account in which multiple lower-level processes contribute to the sense of agency, while a more general monitoring system evaluates their outputs.
CAVEATS: Details about sample size, participant characteristics, and precise task parameters are not specified here. The assumption that decreases in control isolate error detection and increases isolate regularity detection is a theoretical simplification and may not fully capture the underlying processes. The task itself is highly simplified, involving control of a single moving dot, which may not generalize to more complex real-world actions. The study also relies on behavioral and confidence measures without additional neural or physiological data.
KEY TAKEAWAYS: Participants were more sensitive to losing control than to gaining it, indicating that different types of control changes are not equally detectable. However, their ability to judge how well they were performing remained stable across both conditions. This suggests that while the underlying perceptual processes differ, the system used to evaluate performance operates consistently across them.
Recurrent pattern completion supports visual inference in mouse V1
Johnson, T. (2025). Recurrent pattern completion drives the neocortical representation of sensory inference. Nature Neuroscience. doi.org/10.1038/s41593-025-02055-5
TAGS: sensory inference, visual cortex, illusory contours, pattern completion, top-down processing, predictive coding, mesoscale imaging, optogenetics
OVERVIEW: This study examined how the brain infers complete visual structure from incomplete input using mice exposed to illusory contour stimuli. These are images where edges or shapes are perceived even though they are not physically drawn, requiring the brain to “fill in” missing information. The researchers combined multiple recording and manipulation techniques, including electrophysiology and two-photon imaging, to observe and influence activity in primary visual cortex (V1). They identified a subset of neurons that responded selectively to the inferred shape rather than to the individual visual elements that suggested it. When these neurons were artificially activated, they reproduced the same population-level activity pattern seen during perception of the illusory shape, even when no visual stimulus was present.
OF NOTE: The key point is that some neurons in early visual cortex are not just responding to what is directly seen, but to what the system infers is present. Pattern completion refers to the process by which the brain fills in missing information based on prior structure. In this case, recurrent connections within V1 appear to reinforce a pattern corresponding to a complete shape, even when only fragments are available. The fact that activating these neurons alone can recreate the pattern suggests that the representation is internally generated, not just driven by incoming sensory data. This is consistent with models of perception in which the brain continuously combines sensory input with prior expectations to construct what is experienced.
CAVEATS: The study was conducted in mice, and the extent to which these mechanisms generalize to human perception is not specified. Details about sample size, behavioral context, and recording conditions are not provided here. The identification of illusory contour–responsive neurons depends on specific stimuli and analysis methods. The relationship between these neural patterns and subjective perceptual experience is inferred rather than directly measured.
KEY TAKEAWAYS: Early visual cortex contains neurons that encode inferred structure rather than just raw input. These neurons can generate a representation of a complete shape even in the absence of visual input, supporting a role for recurrent pattern completion in perception. This suggests that what is perceived is shaped not only by incoming data but by internally generated structure that organizes incomplete input into coherent forms.
Spontaneous behavior as a succession of self-directed tasks in mice
Weinreb, C., et al. (2026). Spontaneous behavior is a succession of self-directed tasks. Neuron, 114(5), 922–937.e12. doi.org/10.1016/j.neuron.2025.11.021
TAGS: spontaneous behavior, prefrontal cortex, computational ethology, behavioral states, affordances, calcium imaging, striatum, hierarchical modeling
OVERVIEW: This study examined whether spontaneous behavior in freely exploring mice is structured rather than random. Using motion tracking and an unsupervised modeling approach, the authors broke behavior down into fast, repeatable units (“syllables”) and then identified slower, longer-lasting states that persisted for seconds to about a minute. These states corresponded to activities such as wall-following, grooming, or investigating objects, and varied depending on the environment. Neural recordings from dorsomedial prefrontal cortex (dmPFC) and dorsal striatum, along with lesion experiments, linked these behavioral states to distinct patterns of brain activity.
OF NOTE: The important shift here is from viewing spontaneous behavior as unstructured to viewing it as organized into self-directed, task-like states. These states are not externally assigned but arise from the interaction between the animal’s body, its environment, and its neural systems. The concept of affordances is central: the environment offers certain possible actions, and behavior organizes around those possibilities. dmPFC activity tracked which state the animal was in and emphasized information relevant to that state, such as spatial layout or object features. This suggests that prefrontal cortex may help maintain a broader, slower-changing representation of “what is currently being done,” while other regions handle detailed movement.
CAVEATS: The findings are based on mice in controlled laboratory environments, and it is not specified how directly this structure generalizes to more complex or natural settings. The identification of behavioral states depends on modeling assumptions. dmPFC activity often followed state transitions, which limits claims about direct control. Lesions altered behavior but did not eliminate it, indicating that multiple systems contribute.
KEY TAKEAWAYS: Spontaneous behavior is organized into structured, longer-lasting states that function like self-directed tasks. These states depend on interaction with the environment and are supported by distributed neural systems, with prefrontal cortex tracking broader state-level organization and other regions handling detailed action.
Large-scale brain waves dominate cortical activity from slow to fast rhythms
Alexander, D. M., & Dugué, L. (2026). The dominance of large-scale phase dynamics in human cortex, from delta to gamma. eLife. elifesciences.org/articles/100674
TAGS: cortical phase dynamics, traveling waves, stereotactic EEG, large-scale coordination, spatial frequency, gamma oscillations, human cortex
OVERVIEW: This study examined whether cortical activity is primarily organized at small, local scales or across large, spatially extended patterns. The authors reanalyzed stereotactic EEG recordings from 23 epilepsy patients performing a delayed free recall task, using between 31 and 108 intracranial electrodes per participant. They focused specifically on phase, which refers to the timing alignment of oscillatory activity across different brain regions, rather than amplitude or signal strength. Across 34 frequency bands spanning 1 to 97 Hz, they used singular value decomposition and spatial-frequency analysis to estimate how much activity was organized into long-wavelength (large-scale) versus short-wavelength (local) patterns over distances up to approximately 8 to 16 cm.
OF NOTE: The central result is that most of the organized activity was carried by large-scale phase patterns, with power decreasing steadily as spatial scale became finer. This held not only for slower oscillations, where large-scale coordination is expected, but also for higher-frequency activity such as gamma, which is often assumed to reflect more local processing. Phase, in this context, captures how different regions are coordinated in time, even if their activity levels differ. The dominance of long-wavelength phase patterns suggests that much of what appears as local activity is better understood as participation in broader, distributed dynamics. One way to interpret this is that cortical processing is not primarily a collection of independent local computations, but a system in which large-scale coordination provides the temporal structure within which local activity unfolds.
CAVEATS: The data come from epilepsy patients undergoing clinical monitoring, with electrode placement determined by medical need rather than uniform sampling, resulting in sparse and uneven spatial coverage. Spatial scales larger than the recording array were inferred through modeling rather than directly observed. The analysis focused on phase and did not incorporate amplitude, which may carry additional information. The study design does not link specific cognitive processes or task conditions to particular spatial patterns, and interpretation of higher-frequency results requires caution due to potential contamination from lower-frequency components.
KEY TAKEAWAYS: Across the recorded cortical regions, large-scale phase coordination dominates the structure of brain activity from slow to fast frequencies. Local signals largely reflect engagement in these broader patterns rather than isolated, site-specific processing. This supports a view of cortical function as fundamentally distributed, where coordination across regions is a primary organizing feature rather than a secondary effect.
Collective social niche construction shaping adaptive social networks
Sueur, C. (2026). Collective social niche construction shaping adaptive social networks. Trends in Ecology & Evolution. doi.org/10.1016/j.tree.2025.03.010
TAGS: social networks, niche construction, behavioural plasticity, self-organisation, phase transitions, cultural evolution, collective intelligence, network morphology
OVERVIEW: This perspective article synthesizes research across behavioural ecology, network science, and cultural evolution to argue that social networks should be understood as dynamic, adaptive systems shaped through ongoing interaction. Rather than treating networks as static maps of relationships, the paper focuses on how individual behaviors such as attraction, avoidance, grooming, and partner choice accumulate over time to produce network structures. These structures, in turn, influence processes such as information flow, cooperation, and disease transmission. The author introduces a conceptual framework linking behavioral plasticity, cognitive capacity, and the ability of networks to reorganize in response to changing conditions.
OF NOTE: A key concept is collective social niche construction, which extends the idea that organisms modify their environments to include the social environment itself. In this view, individuals do not just occupy a network but actively shape it through their interactions. These local interactions can generate emergent network properties such as clustering, modularity, and efficiency without centralized control. The paper also discusses phase transitions, where small changes in interaction patterns can produce large, qualitative shifts in network structure. For example, slight changes in who interacts with whom can reorganize a network from loosely connected to highly clustered. These structural changes then feed back to influence individual behavior, shaping what information is available, how quickly it spreads, and which forms of coordination are possible.
CAVEATS: The article is a conceptual synthesis and does not present new empirical data, formal models, or quantitative tests of the proposed framework. Examples are drawn from a wide range of species and contexts, making it difficult to assess the strength of evidence for specific mechanisms. Key constructs such as network adaptability and collective social niche construction are described but not operationalized with standardized metrics, limiting immediate testability. Potential confounds, including environmental variation and measurement limitations in network reconstruction, are not treated in detail.
KEY TAKEAWAYS: Social networks can be understood as adaptive systems that emerge from ongoing interaction rather than as fixed structures. Local behavioral rules can scale into large-scale network organization, and these structures feed back to shape individual behavior and collective outcomes. The framework highlights how coordination, information flow, and resilience depend on the dynamic interplay between individual actions and evolving network structure.
Conclusion
Across these studies, a more specific picture comes into focus. In each case, the system is not simply reacting to inputs, but organizing them into a structured, workable model that guides what can be perceived and how action unfolds. That model is shaped by constraint: what can be detected, what can be integrated, and what can be stabilized long enough to influence behavior.
What varies across the studies is not whether modeling occurs, but how it is limited. In sensorimotor control, the system is more sensitive to breakdown than to improvement, biasing detection toward loss. In perception, incomplete input is actively completed into coherent structure, with internally generated patterns shaping what is seen. In behavior, action does not remain open-ended but settles into sustained, context-sensitive states that organize what the system continues to do. In cortical dynamics, local activity is embedded within large-scale coordination patterns that structure when and how regions participate. In social systems, individual interactions accumulate into network configurations that constrain what information circulates and how coordination can adapt.
Taken together, these findings point to a common condition: systems operate on structured partial information that is treated as sufficient. The resulting model can be coherent, stable, and functional, but it does not represent the full set of conditions available. From within the system, there is no direct indication of what has been excluded, only the experience of a working model that appears complete.
This is where emergence becomes consequential. The capacity for new forms of response depends on whether the system can register and incorporate information that falls outside its current organization. If incoming signals are not detected, not retained, or not allowed to alter the existing structure, the system will continue to reproduce established patterns, even as conditions shift around it.
For readers working in research, clinical, or organizational settings, the implication is critical. Change does not occur simply by introducing new information or alternative options. It depends on whether the system can take that information in and reorganize around it. Where that capacity is limited, behavior can remain coherent and even effective within its frame, while still being constrained by what it cannot register, integrate, or coordinate.
#research | #theory | #practice
About emotive.energy
We study how people think, feel, and respond in real time, and how those patterns interact in groups and systems. As a research-led co-op and working laboratory, we help individuals, teams, and organizations recognize and shift the dynamics that shape experience, including stress, decision-making, relationships, creativity, and meaning. Our work spans contexts from workplaces and leadership to family systems, artistic practice, and personal development. Through coaching, group programs, and proprietary tools, we make these dynamics visible as they form so people can work with them directly. The result is clearer thinking, more effective action, and more adaptive ways of relating under pressure and in everyday life. [website]
Links
Website: emotive.energy
Substack: Posts · Notes · Podcast
Social: Twitter / X · BlueSky · YouTube · Facebook · LinkedIn · Instagram
Podcast: Apple Podcasts · Spotify · Pocket Casts · Overcast