{"id":3163,"date":"2025-12-21T00:01:32","date_gmt":"2025-12-21T00:01:32","guid":{"rendered":"https:\/\/beyondtheimpact.net\/?p=3163"},"modified":"2025-12-21T00:01:32","modified_gmt":"2025-12-21T00:01:32","slug":"the-two-state-brain-and-bidirectional-evidence","status":"publish","type":"post","link":"https:\/\/beyondtheimpact.net\/?p=3163","title":{"rendered":"The two-state brain and bidirectional evidence"},"content":{"rendered":"

<\/a><\/p>\n

The notion of a two-state brain captures the idea that large-scale neural activity often settles into one of two relatively stable regimes, or attractor states, which can correspond to alternative interpretations, choices, or perceptual outcomes. Rather than operating as a smooth continuum of intermediate possibilities, cortical circuitry frequently displays dynamics in which activity patterns abruptly \u201csnap\u201d into one of two dominant configurations. These configurations can be characterized by distinct patterns of firing across neuronal populations, differences in oscillatory synchrony, and shifted balances between excitation and inhibition. In this view, the brain\u2019s dynamics resemble a system navigating an energy landscape: shallow wells allow rapid switching and high sensitivity to noise, whereas deep wells enforce stability and resistance to change. A two-state regime emerges when the energy landscape is effectively dominated by two such wells, and neural activity keeps hopping between them under the influence of internal fluctuations and external inputs.<\/p>\n

At the microscopic level, these dynamics depend on recurrent connectivity, which supports self-sustaining patterns of activity. Neurons that encode similar features tend to be heavily interconnected, so once one subset of neurons becomes active, recurrent excitation amplifies that activity while lateral inhibition suppresses competing populations. This architecture naturally gives rise to winner-take-all behavior and bistability. For example, in a decision circuit comparing two alternatives, neurons coding for option A and option B inhibit each other while exciting themselves. Small, transient differences in input or intrinsic noise become magnified by positive feedback, driving the network into a state where either the A-coding or B-coding population dominates. The system\u2019s trajectory in neural state space thus flows toward one of two attractors, generating a robust two-state pattern even when the initial inputs are ambiguous or noisy.<\/p>\n

Oscillatory coordination across brain regions further shapes these two-state dynamics. Distinct attractor states are often associated with characteristic rhythms and phase relationships, reflecting how neural assemblies synchronize when representing a particular interpretation or decision. For instance, gamma-band synchrony within a local circuit can signal a currently dominant representation, while slower theta or alpha rhythms modulate transitions between attractors by periodically varying excitability. When a specific state becomes dominant, its associated oscillatory pattern can entrain activity across hierarchically higher or lower areas, reinforcing a coherent global configuration. This coordination allows the two-state pattern to propagate across the cortical hierarchy, aligning perception, action planning, and internal evaluation around a consistent neural state.<\/p>\n

The concept of predictive processing and priors provides a functional lens on these neural regimes. Under predictive processing, the brain is continually generating top-down predictions about sensory input and comparing them with bottom-up signals. Strong priors bias the system toward particular interpretations, effectively deepening the corresponding attractor wells in neural state space. When a prior is very strong, neural activity quickly converges on the corresponding state and resists disconfirming evidence. Conversely, weak or uncertain priors flatten the landscape, making the brain more sensitive to sensory evidence and more prone to switching between states. Thus, the two-state brain can be understood as a prediction engine that alternates between competing, self-consistent hypotheses about the world, with priors and prediction errors dynamically reshaping the stability and accessibility of each state.<\/p>\n

At a more abstract level, these dynamics dovetail with the computational framework of bayesian inference. Neural populations encode probabilistic beliefs about hidden causes of sensory inputs, and their activity patterns approximate posterior distributions over these causes. In a two-state regime, the posterior becomes sharply bimodal, concentrating probability mass around two competing hypotheses. The neural system then evolves as if it were sampling from, or committing to, one of these modes. Synaptic weights and neuromodulatory signals determine the effective \u201ctemperature\u201d of this sampling process: high temperature corresponds to noisy wandering between states, while low temperature leads to strongly locked-in choices. This probabilistic perspective clarifies why the same external stimulus can lead to different perceptual outcomes across trials, as the internal bayesian machinery sometimes settles into one mode and sometimes into the other, depending on momentary fluctuations and prior expectations.<\/p>\n

Empirical findings from perceptual bistability offer clear illustrations of two-state dynamics. In phenomena like binocular rivalry or ambiguous figures, neural activity in visual cortex alternates between distinct ensembles corresponding to the competing percepts, even though the physical stimulus remains constant. Imaging and electrophysiology reveal that these alternations are not random flips but reflect structured transitions within a recurrent network. Activity ramps up in one population while declining in its competitor, with cross-inhibition shaping a clean switch. Periods of dominance correspond to the system being trapped in one attractor, whereas transition periods show increased variability and desynchronization, consistent with the system momentarily leaving one basin of attraction and approaching another. Such findings highlight that the two-state brain is not just a conceptual metaphor but is grounded in measurable shifts in large-scale population activity.<\/p>\n

Beyond sensory cortices, higher-order association areas, particularly in prefrontal and parietal regions, show analogous bistable dynamics during complex decisions and working memory tasks. When a subject holds one of two rules in mind, or commits to a specific intention, neural ensembles in these regions exhibit stable activity patterns that can persist in the absence of ongoing input. These stable patterns can be rapidly reconfigured when new information or task demands arise, effectively flipping the system into the alternate state. The ability to maintain a current state while remaining ready to switch underlines the balance between robustness and flexibility that characterizes the two-state brain. Neuromodulators such as dopamine and norepinephrine modulate this balance by altering gain and plasticity, thereby controlling how easily the brain transitions between alternative attractors.<\/p>\n

These neural dynamics also intersect with theories of consciousness that emphasize global integration and access. One family of ideas proposes that conscious experience corresponds to brain states in which information is globally broadcast across widely distributed networks. In a two-state framework, one attractor may correspond to a globally integrated configuration that supports conscious access to a particular interpretation or decision, while the alternative state might reflect a competing, potentially preconscious or suppressed configuration. The transitions between such states can be influenced by attention, task relevance, and learning history, which determine which representations get promoted to global availability. This linkage suggests that shifts in conscious perception may be underpinned by the same kind of bistable dynamics that govern simpler sensory or motor choices.<\/p>\n

Noise and variability, far from being mere nuisance factors, play a constitutive role in enabling the two-state brain to explore its dynamic repertoire. Stochastic fluctuations in synaptic transmission, spontaneous firing, and external disturbances create perturbations that can push the neural system across the boundary between attractors. If the system were perfectly deterministic and noise-free, it might become stuck in a single dominant state and lose adaptability. Instead, controlled levels of noise, filtered through the structure of the recurrent network, allow the system to occasionally escape from one configuration and discover another, particularly when environmental contingencies change. In this light, noise is not simply random error but a resource that supports flexible switching and prevents pathological over-stability.<\/p>\n

Development and learning shape the parameters of the two-state brain by sculpting synaptic strengths and connectivity motifs. Repeated exposure to particular patterns of stimuli or tasks deepens certain attractors through Hebbian plasticity, making those states easier to enter and harder to leave. Conversely, lack of use or active suppression can shallow or even eliminate attractors associated with disfavored interpretations or actions. Over time, this process tunes the brain\u2019s dynamical landscape to the statistical structure of the environment, aligning neural states with relevant categories, goals, and behavioral repertoires. Plastic changes in inhibitory circuits, in particular, fine-tune the competitive interactions that define the boundaries between attractors, directly influencing how sharply the brain operates as a two-state system versus a more graded, multi-stable one.<\/p>\n

Pathological conditions provide additional insight into the neural dynamics of the two-state brain. In certain psychiatric and neurological disorders, attractor states may become excessively deep or abnormally shallow. For example, overly stable attractors could manifest as rigid beliefs, perseverative thoughts, or compulsive behaviors that resist updating despite new evidence, while overly labile dynamics might present as distractibility, unstable perception, or difficulty maintaining a consistent goal. Disruptions in excitation\u2013inhibition balance, synaptic plasticity, or neuromodulatory tone can all tilt the system toward one or the other extreme. Understanding these disorders through the lens of two-state dynamics motivates interventions\u2014pharmacological, behavioral, or neuromodulatory\u2014that aim to rebalance attractor depth and facilitate appropriate transitions between neural states.<\/p>\n

Mechanisms of bidirectional evidence accumulation<\/h3>\n

Bidirectional evidence accumulation refers to the brain\u2019s ability to integrate information not only from past sensory inputs forward in time, but also from anticipated future outcomes and goals backward onto earlier processing stages. In a two-state regime, this means that both bottom-up sensory signals and top-down predictions act on the same competing attractor states, constantly revising their relative stability. Rather than a one-way pipeline in which evidence simply piles up until a decision bound is crossed, the system behaves as a recurrent loop: early interpretations bias the sampling of new evidence, and later, more abstract evaluations feed back to reweight earlier sensory or mnemonic traces. This looping structure naturally produces a bidirectional flow of influence even though the underlying physiology is locally causal in time.<\/p>\n

Within the framework of bayesian inference, bidirectional accumulation can be understood as a continual updating of posterior beliefs through iterative exchanges between levels of the cortical hierarchy. Sensory areas compute likelihoods based on incoming stimuli, while higher areas provide priors shaped by expectations, context, and task demands. Each attractor state encodes a hypothesis about the hidden causes of sensory input, and its \u201cpull\u201d on the system reflects the current posterior probability of that hypothesis. As new data arrive, prediction errors propagate upward, adjusting these posteriors, while updated priors propagate downward, reshaping how future prediction errors are interpreted. The result is that evidence flows in both directions: sensory evidence refines priors, and priors condition the interpretation of subsequent evidence, altering the odds that the system will settle into one attractor versus the other.<\/p>\n

Predictive processing provides a more mechanistic account of how this bidirectional exchange unfolds in neural circuits. In predictive processing, higher-level populations generate predictions about lower-level activity, and lower levels signal deviations from these predictions as error signals. In a two-state brain, each of the competing states corresponds to a distinct generative model of the world, with its own predicted pattern of sensory input and internal context. When one model dominates, its predictions suppress alternative patterns by reducing their gain and attenuating errors that would otherwise favor the competing state. However, if the incoming evidence persistently contradicts the dominant state\u2019s predictions, prediction errors accumulate, gradually eroding that state\u2019s stability while bolstering its rival. This process is inherently bidirectional: predictions flow down, errors flow up, and the balance between the two attractors is continually renegotiated by their success or failure in explaining the sensory stream.<\/p>\n

The notion of priors is crucial for understanding why evidence accumulation in a two-state system is not symmetric. Priors effectively serve as initial conditions and ongoing boundary conditions that bias the trajectory of neural activity. A strong prior for one state deepens its attractor well, so that even modest confirmatory evidence is enough to maintain dominance, while contradictory evidence must be unusually strong or persistent to trigger switching. Conversely, when priors are weak or evenly balanced, even small pieces of evidence can tilt the system from one state to the other. This asymmetry means that the same sensory input can drive very different neural trajectories depending on the prior configuration, and that the direction of evidence flow\u2014the degree to which new data can push the system \u201cback\u201d toward a less favored state\u2014is itself a function of learned expectations and contextual framing.<\/p>\n

Bidirectional accumulation becomes especially apparent when considering how later, more integrative evaluations feed back to reshape earlier neural representations. For instance, when a putative choice is evaluated against internal goals and long-term consequences in prefrontal circuits, the resulting value signals can alter the effective gain of sensory or premotor representations supporting each alternative. If the downstream valuation network strongly favors one outcome, it can retroactively amplify earlier evidence in its favor by enhancing the activity of neurons encoding that option while suppressing competing populations. From a dynamical systems viewpoint, this constitutes a backward-acting force that reconfigures the energy landscape after initial evidence has already begun to push the system toward one attractor. Although no true retrocausality is involved\u2014signals still propagate forward in physical time\u2014the functional effect is that later evaluations reshape the interpretation and weight of earlier evidence.<\/p>\n

Local circuit mechanisms implement this bidirectional shaping through recurrent excitation, cross-inhibition, and top-down modulation of synaptic efficacy. In a simple two-pool decision circuit, each pool receives both feedforward sensory input and feedback signals from higher areas encoding context, rules, or value. As sensory evidence accumulates, recurrent excitation within a pool amplifies small differences, while mutual inhibition drives the system toward a winner-take-all outcome. Simultaneously, feedback projections modulate the baseline excitability and effective gain of each pool. A top-down bias can, for example, elevate the resting activity of one pool, making it more responsive to incoming spikes and effectively \u201cpre-accumulating\u201d evidence in its favor before the stimulus even arrives. During the course of a trial, evolving top-down signals can strengthen or weaken these biases, effectively revising the accumulated evidence without needing to overwrite or erase earlier spikes.<\/p>\n

Temporal integration in such circuits is governed not only by how spikes are added but also by how they are discounted over time. Short-term synaptic plasticity, adaptation, and changes in membrane time constants determine how long past inputs continue to contribute to the present state. Bidirectional evidence accumulation arises when top-down influences dynamically adjust these temporal windows. For instance, neuromodulatory input can prolong the integration window for one pool while shortening it for the other, causing the network to \u201cremember\u201d favorable evidence for one choice longer than for its competitor. This selective retention introduces a directional tilt: evidence supporting the privileged state is effectively integrated over a longer horizon, whereas disconfirming evidence decays more rapidly. As conditions evolve, the system can reverse this tilt, retroactively down-weighting previously favored evidence by accelerating its decay.<\/p>\n

Cross-level interactions between sensory, associative, and motor areas further enhance bidirectionality by coupling evidence accumulation across multiple representational domains. When a high-level hypothesis about a scene or task becomes active, it not only biases the interpretation of low-level features but also constrains the set of actions considered appropriate. Feedback from motor planning areas can then influence sensory processing by prioritizing features relevant to the planned action, effectively filtering new evidence through the lens of the emerging decision. This closed-loop arrangement ensures that evidence accumulation at one level is continuously shaped by the current state of other levels, so that revisions at a downstream stage can propagate backward to adjust upstream representations, even while new evidence continues to flow forward.<\/p>\n

Oscillatory dynamics provide a temporal scaffold for these bidirectional flows. Different frequency bands can gate feedforward versus feedback communication, with higher frequencies often associated with bottom-up transmission and lower frequencies with top-down influence. In a two-state network, dominance of one attractor may be accompanied by a particular pattern of phase coupling that favors either feedforward or feedback channels. During periods when top-down influence is dominant, the system effectively \u201cre-reads\u201d its recent history, allowing higher areas to re-evaluate and re-weight earlier activity patterns in lower areas. Conversely, when feedforward channels dominate, fresh sensory evidence can penetrate and challenge the currently prevailing state. Switching between these oscillatory regimes enables the brain to alternate between consolidating a current interpretation and opening itself to revision, thereby orchestrating the bidirectional accumulation of evidence over time.<\/p>\n

On a more abstract level, bidirectional accumulation in a two-state brain can be framed in terms of iterative message-passing algorithms used in approximate bayesian inference. Each iteration involves exchanging \u201cmessages\u201d that encode beliefs about hidden variables between connected nodes in a graphical model. In neural terms, these messages are patterns of firing that convey predictions, likelihoods, and error signals between populations. Because messages are updated based on both incoming and outgoing signals, each population\u2019s state reflects an ongoing compromise between its own local evidence and the constraints imposed by its neighbors. Over successive cycles, this process can converge toward a stable assignment of beliefs corresponding to one of the two attractor states. During early iterations, however, the network may oscillate or vacillate between tentative assignments, as new messages retroactively alter the interpretation of prior ones, capturing the essence of bidirectionality in evidence accumulation.<\/p>\n

In perceptual bistability, the same physical stimulus supports two distinct interpretations, and the system alternates between them over time. Bidirectional accumulation provides a natural explanation for how these alternations arise without any change in the external input. Internal noise, adaptation, and fluctuating top-down biases continuously perturb the balance of evidence supporting each interpretation. When the currently dominant interpretation starts to generate mounting prediction errors\u2014because adaptation has weakened the neural populations that instantiate its predictions, or because higher-level systems begin to favor a different global configuration\u2014feedback signals gradually erode its attractor depth. At the same time, small, previously subthreshold signals favoring the alternative interpretation become more impactful, and their influence on downstream evaluative circuits grows. This interplay of forward and backward influence allows the system to drift toward the competing state, culminating in a perceptual switch.<\/p>\n

The same principles extend to decisions involving abstract beliefs and goals, where evidence may come not only from external stimuli but also from internal simulations, memories, and imagined futures. Higher-order areas can generate counterfactual scenarios and evaluate their plausibility, feeding the resulting evidence back into mid-level and sensory representations. In a two-state deliberation\u2014for example, accepting versus rejecting a proposition\u2014one state may be supported by currently perceived facts, while the other is buttressed by anticipated consequences or remembered outcomes. As these internally generated pieces of evidence accumulate, they can retroactively change how earlier observations are categorized and weighted. Thus, the brain\u2019s capacity for imagination and prospection injects a temporal richness into evidence accumulation, where what \u201ccounts\u201d as evidence at one moment can be reinterpreted at a later moment in light of newly simulated possibilities.<\/p>\n

Switching thresholds and state transitions<\/h3>\n

In a two-state brain, switching does not occur continuously but rather when neural activity crosses specific thresholds that delineate one attractor basin from another. These switching thresholds can be conceptualized as boundaries in a high-dimensional state space, beyond which the self-reinforcing dynamics of one state give way to those of the competing state. Within the basin of an attractor, recurrent excitation and local stabilization mechanisms keep fluctuations contained; only when perturbations grow large enough, or when the basin itself becomes shallower, does the system escape. This escape often manifests as a rapid, non-linear transition: neural activity that had been fluctuating around one stable pattern suddenly diverges and converges on the alternative pattern, producing the phenomenological impression of an abrupt change in perception, thought, or intention.<\/p>\n

Thresholds for switching are not fixed physical barriers but emergent properties of the underlying circuitry and neuromodulatory environment. The same network can exhibit different effective thresholds depending on synaptic strengths, membrane excitability, and the current balance of inhibition and excitation. For example, a high level of global inhibition can raise the effective threshold by dampening perturbations, making it harder for noise or conflicting input to push the system out of a dominant state. Conversely, reduced inhibition or increased excitatory gain can lower the threshold, allowing small changes in input or intrinsic fluctuations to trigger transitions. In computational terms, these parameters adjust the curvature of the energy landscape, with steeper walls reinforcing state stability and flatter walls facilitating transitions across boundaries.<\/p>\n

Predictive processing and priors play a pivotal role in shaping where these thresholds lie and how easily they are crossed. Strong priors associated with one hypothesis deepen its attractor and effectively move the switching boundary closer to the competing state, demanding more cumulative contradictory evidence before a transition can occur. Under such conditions, prediction errors must build up over time to a critical level before they can overcome the stabilizing influence of expectations. When priors are weak or uncertain, prediction errors have a greater relative impact, lowering the effective threshold and increasing the frequency of state transitions. This relationship between priors and switching thresholds means that adaptive behavior relies on correctly calibrating how resistant current states should be to revision, depending on environmental volatility and task demands.<\/p>\n

Mechanistically, thresholds can be realized through non-linear transfer functions at the level of single neurons and microcircuits. Many neurons exhibit sigmoidal input\u2013output relationships, in which firing rate remains low until synaptic input passes a critical level, after which firing rapidly ramps up. When such neurons are embedded in recurrent loops, their collective responses sharpen into quasi-discrete shifts between low- and high-activity states. Cross-inhibitory motifs further accentuate this behavior: when one population\u2019s activity grows past a threshold, it suppresses its competitor more strongly, creating a positive feedback loop that accelerates the transition. The effective threshold is thus the point at which the net recurrent feedback, including inhibition of the alternative state, becomes self-sustaining and drives the network fully into the new configuration.<\/p>\n

Temporal integration parameters critically determine how quickly thresholds are approached. Neural systems that integrate inputs over long timescales allow weak but persistent evidence to accumulate, gradually nudging the system toward the switching boundary. In contrast, short integration windows favor responsiveness to rapid, high-amplitude changes, but discount slow drifts. Adaptation mechanisms, such as synaptic depression or spike-frequency adaptation, gradually diminish the responsiveness of neurons that have been active for a prolonged period, effectively weakening the stabilizing forces holding the current state. This slow erosion can bring the system closer to the switching threshold even in the absence of dramatic new evidence, setting the stage for a transition triggered by modest fluctuations that would previously have been insufficient.<\/p>\n

Oscillatory rhythms add a rhythmic modulation to switching thresholds by periodically altering neuronal excitability and effective connectivity. During specific phases of a theta or alpha cycle, neuronal populations may be more or less responsive to incoming input. As a result, the same perturbation may push the system across a threshold if it occurs at a high-excitability phase, but fail to do so at a low-excitability phase. Gamma-band synchronization can transiently increase the coherence of a dominant attractor, briefly raising the threshold for switching by aligning firing across the supporting ensemble. Conversely, desynchronization periods correspond to moments when the system is more labile, thresholds are effectively lower, and transitions become more likely. Thus, rhythmic fluctuations in network synchronization create temporal windows in which state transitions are facilitated or inhibited.<\/p>\n

Switching thresholds are also modulated by higher-order control systems that regulate cognitive flexibility. Prefrontal and cingulate regions, often associated with monitoring conflict and error, can adjust the gain and inhibitory tone in downstream circuits that implement two-state dynamics. When conflict between competing representations is high, or when prediction errors accumulate, these control systems can lower thresholds by boosting noise levels, reducing inhibition, or enhancing the influence of alternative hypotheses. This proactive adjustment makes it easier for the system to leave entrenched states when they are no longer adaptive. Conversely, when stability is prioritized\u2014for instance, during sustained attention or goal maintenance\u2014control systems can raise thresholds, making spontaneous transitions rarer and enforcing persistence in the current attractor.<\/p>\n

In the framework of bayesian inference, state transitions can be viewed as moments when the posterior distribution shifts its mass from one mode to another in response to accumulating evidence. The switching threshold corresponds to the point at which the posterior probability of the alternative hypothesis surpasses that of the currently dominant one, or at least becomes sufficiently close that stochastic fluctuations can tip the balance. Neural implementations approximate this comparison through competing population codes: firing rates in each population represent log probabilities or log odds, and a transition occurs when the difference in activity reverses sign or falls below a critical margin. Intrinsic noise and limited sampling ensure that transitions do not occur only at a single precise value, but within a probabilistic band around this decision boundary.<\/p>\n

At the behavioral level, these neural thresholds manifest as psychophysical decision boundaries, such as the criterion in signal detection tasks or the decision bounds in drift\u2013diffusion models. In a two-state drift\u2013diffusion scenario, evidence for one alternative accumulates until it reaches a bound, triggering a choice. The height of this bound maps onto the neural switching threshold: high bounds correspond to deep attractors and strong resistance to switching, whereas low bounds reflect shallow attractors and a willingness to accept decisions based on less evidence. Adjustments in bound height seen experimentally\u2014for example, when participants are instructed to favor speed over accuracy\u2014mirror changes in underlying neural thresholds implemented via gain control, neuromodulation, and alterations in recurrent connectivity.<\/p>\n

State transitions often unfold in a stereotyped sequence, revealing intermediate metastable configurations as the system moves from one attractor to another. During these excursions, neural population trajectories can follow low-dimensional paths through state space, as revealed by techniques like principal component analysis or manifold learning applied to population recordings. The initial phase of the transition typically involves growing variability and partial disengagement from the old state, followed by a brief period of high-dimensional exploration, and then convergence into the new state\u2019s characteristic pattern. The detailed shape of this trajectory depends on the geometry of the state space and on how thresholds are implemented; for example, the system may pass near saddle points that act as transient waystations, influencing the timing and reliability of the switch.<\/p>\n

Importantly, thresholds do not merely govern transitions between perceptual or cognitive states; they also define when neural changes become accessible to consciousness. Many fluctuations in subthreshold activity may reflect latent competition that never crosses the boundary required for global broadcasting. Only when activity supporting an alternative interpretation exceeds a critical threshold does it recruit long-range synchronization and ignite widespread cortical networks, making the new state available for report and voluntary control. In this view, transitions in consciousness correspond to moments when internal dynamics cross a global access threshold, rather than continuous tracking of all subthreshold variations in neural activity. This perspective links microcircuit-level switching phenomena to macroscopic shifts in what is consciously experienced.<\/p>\n

Pathological alterations in switching thresholds can produce distinctive clinical profiles. Excessively high thresholds may yield inflexible cognition and perception, where individuals remain locked into particular interpretations or behavioral patterns despite accumulating contradictory evidence. Such rigidity can appear in disorders characterized by perseveration, rumination, or delusional conviction, where priors are so strong and attractors so deep that prediction errors rarely suffice to trigger transitions. On the other hand, abnormally low thresholds can generate instability and distractibility, as the system switches states too frequently in response to minor fluctuations. This can manifest as rapid shifts in mood, attention, or perceptual organization. Therapeutic interventions that target neuromodulatory systems or specific circuit parameters can, in principle, rescale these thresholds, restoring a more balanced pattern of state transitions.<\/p>\n

Learning continually recalibrates switching thresholds through experience-dependent plasticity. When particular transitions consistently lead to successful outcomes, the system can lower the threshold for those transitions by strengthening pathways that support them or by weakening stabilization of the current state. Conversely, transitions that yield negative outcomes can be discouraged by deepening the current attractor or by enhancing inhibitory control over the alternative. Over time, this reinforcement-driven sculpting produces individualized profiles of readiness to switch in particular domains: some decisions become nearly automatic and require minimal evidence to trigger transitions, while others demand substantial and prolonged conflict before the system abandons its current state. This dynamic tuning ensures that switching thresholds remain aligned with the statistical structure of the environment and with the organism\u2019s goals, maintaining an adaptive balance between stability and flexibility in the two-state brain.<\/p>\n

Implications for decision-making and perception<\/h3>\n

The two-state perspective casts decision-making as a sequence of commitments and reversals between discrete, self-stabilizing neural configurations rather than as a smooth, continuously graded adjustment of preferences. When the brain evaluates alternatives, populations encoding each option compete to recruit a global pattern of activity that coordinates sensory, associative, and motor regions. Once one option\u2019s attractor has gained dominance, ongoing processing is filtered through that state: sensory features consistent with the chosen interpretation are preferentially amplified, whereas conflicting features are suppressed or reinterpreted. This feedback steadily increases the coherence of the chosen state, making it progressively harder for competing alternatives to gain a foothold unless accumulating contradictions or new goals erode the attractor\u2019s depth. As a result, the temporal profile of a decision is not simply a linear weighting of evidence but a path-dependent trajectory through the underlying dynamical landscape.<\/p>\n

Within the framework of bayesian inference, these dynamics embody a concrete neural implementation of prior\u2013likelihood\u2013posterior interactions. Each attractor corresponds to a hypothesis about the environment or about one\u2019s own actions, with its stability reflecting the joint influence of sensory likelihoods and top-down priors. Strong priors for a given hypothesis deepen its associated attractor, biasing the decision process toward that outcome even when evidence is mixed. This is especially apparent in ambiguous or noisy situations, where the external input underdetermines the correct interpretation. Under such conditions, the state favored by existing expectations tends to capture the system first, and only sustained or unusually strong disconfirming evidence can drive a transition to the rival attractor. In this way, the balance between predictive processing and priors on the one hand, and ongoing sensory input on the other, shapes which decisions are reached and how resistant they are to revision.<\/p>\n

Perception is likewise governed by attractor competition, such that conscious experience usually reflects whichever representational state has successfully recruited the largest and most integrated coalition of neural assemblies. When one percept or interpretation dominates, it not only occupies relevant sensory areas but also enlists higher-order association networks, allowing that state to influence attention, working memory, and action planning. This widespread recruitment yields a sense of continuity and coherence: despite noisy and fragmentary input, the system rapidly settles into a consistent percept that can guide behavior. However, the same mechanism also explains why perception can be strikingly discontinuous. At moments when prediction errors accumulate or contextual cues shift, the currently dominant attractor may lose stability, precipitating an abrupt reorganization of neural activity into an alternative configuration. Subjectively, this manifests as a sudden change in what is seen, heard, or understood, even though the physical input may have changed only gradually\u2014or not at all.<\/p>\n

Ambiguous stimuli offer a window into how decision-making and perception interlock in a two-state brain. In binocular rivalry, for instance, each eye receives a different image, and the percept alternates between them rather than blending. The perceptually dominant image corresponds to one attractor state, while the suppressed image is represented weakly or subcortically in the competing state. Decision-like processes in higher areas continually monitor prediction errors, contextual relevance, and task demands, adjusting thresholds for switching. When the suppressed representation gradually gains strength due to adaptation in the dominant network or to shifts in attention, the system eventually crosses a switching boundary, and the percept flips. Here, the line between \u201cperceptual\u201d and \u201cdecisional\u201d is blurred: the same competitive architecture that supports overt choices also shapes which of two potential percepts becomes consciously available at any given moment.<\/p>\n

The linkage between attractor dominance and consciousness can be formalized by viewing global availability as a secondary threshold layered on top of local two-state competition. Many neural configurations may arise transiently in early sensory cortices without ever becoming the content of experience because they fail to achieve the degree of integration and reinforcement necessary to ignite a large-scale network. Only when a particular configuration recruits enough recurrent support\u2014both within its local circuit and across distant regions\u2014does it cross a global access threshold and become reportable. In this sense, consciousness tracks not every fluctuation between candidate states but only those transitions that reach a critical level of coherence. The two-state brain thus supports a hierarchy of selections: local competitions determine which representations are viable candidates, and higher-level, more distributed dynamics decide which candidate wins access to conscious control and long-term influence on behavior.<\/p>\n

This hierarchical selection has direct implications for how biases and expectations shape both what is perceived and the decisions that follow. Because priors deepen some attractors more than others, they make certain percepts easier to ignite to the level of consciousness and certain decisions easier to commit to. For example, in a noisy auditory environment, speech sounds consistent with a listener\u2019s language and expectations are more likely to dominate attractor dynamics in auditory and language areas, leading to a stable percept of meaningful speech rather than to fragmented noise. At the same time, decision circuits that evaluate possible interpretations or responses rely on these biased perceptual states as their input, reinforcing a cycle in which initial expectations guide perception, which then supplies \u201cevidence\u201d that appears to confirm those expectations. In domains like social judgment or risk assessment, this feedback loop can yield stable but potentially distorted patterns of belief and choice.<\/p>\n

Bidirectional evidence accumulation accentuates these effects by allowing later evaluative and motivational signals to reshape earlier processing. As prefrontal and limbic circuits compute value and goal relevance, they send feedback to sensory and association areas, selectively amplifying or attenuating representations that align with current aims. In a two-state regime, this means that a tentative decision or desired outcome can preemptively stabilize the corresponding attractor, making confirmatory cues appear more salient and contradictions easier to ignore. Over time, the network\u2019s trajectory becomes increasingly constrained by the emerging decision, such that alternative outcomes are not just less favored but less fully represented at all. This mechanism facilitates efficient, goal-focused behavior but can also produce confirmation bias and post hoc rationalization, as the system retrofits its perceptual and mnemonic states to support the attractor that has already gained dominance.<\/p>\n

Speed\u2013accuracy trade-offs in decision-making reflect deliberate control over the depth and width of attractor basins. When rapid responses are prioritized, control systems effectively lower decision bounds, making it easier for transient fluctuations in activity to push the network into a committed state. Neural correlates of this adjustment include reduced inhibition, shortened integration windows, and increased gain, all of which shallow competing basins and accelerate transitions. The cost is a higher probability that noise or misleading early evidence will drive the system into the wrong attractor. When accuracy is prioritized, the opposite adjustments occur: integration windows lengthen, inhibition increases, and attractors deepen. The network then requires more consistent evidence to switch states, yielding slower but more reliable decisions. Thus, classic behavioral phenomena in choice tasks arise naturally from the tuning of two-state dynamics rather than from separate mechanisms.<\/p>\n

In sequential decision contexts, the two-state architecture implies a strong dependence of current choices on recent neural history, even when the task structure does not objectively warrant such dependence. After the system has occupied a particular attractor, synaptic and cellular adaptation can temporarily lower its depth, making a switch to the alternative state more likely on the next trial. Conversely, reinforcement or successful outcomes can deepen the recently visited attractor, biasing the system toward repeating the same choice even when evidence is neutral. These carryover effects produce serial dependencies in behavior\u2014such as post-error slowing or choice repetition biases\u2014that mirror the dynamics of how attractor basins are reshaped by experience. From the standpoint of bayesian inference, the system is updating not just beliefs about the external world but also meta-beliefs about the reliability of particular internal states, and these updates feed back to influence subsequent competition between attractors.<\/p>\n

Perceptual learning can be understood as a long-term reconfiguration of the attractor landscape, altering both the granularity and the thresholds of state transitions. As an organism becomes more skilled at discriminating similar stimuli, previously overlapping or weakly separated attractors in sensory cortex become more distinct, enabling faster and more reliable switches between them. This sharpening allows the brain to carve the continuous variability of sensory input into more precisely defined perceptual categories, each associated with its own basin of attraction. Decisions that depend on those categories then inherit this improved separation: the relevant decision circuits receive clearer, more stable inputs, reducing ambiguity and response variability. However, the same mechanisms can lock in maladaptive categories or stereotypes if learning disproportionately deepens attractors corresponding to biased interpretations of ambiguous cues.<\/p>\n

Clinical phenomena such as compulsions, phobias, and persistent negative beliefs highlight how decision-making and perception can become trapped in pathological attractor configurations. In an anxiety disorder, for instance, threat-related interpretations may occupy unusually deep basins, so that even mildly ambiguous stimuli drive the system into a \u201cdanger\u201d state that is hard to exit. Perceptual systems then tune to threat-consistent features, and decision circuits favor avoidance or hypervigilant responses, further reinforcing the dominance of the fear-related attractor. Conversely, conditions involving impulsivity and distractibility may feature overly shallow attractors and low switching thresholds, so that small fluctuations in input or internal noise suffice to trigger frequent state changes. This yields unstable perception and rapidly shifting choices that appear erratic or poorly grounded in evidence. Interventions that modulate neuromodulatory tone, alter inhibitory balance, or train alternative cognitive strategies can, in principle, reshape the attractor landscape, restoring a more adaptive balance between stability and flexibility in both perception and decision.<\/p>\n

At the level of subjective experience, two-state dynamics help explain why decisions often feel categorical and why changes of mind can be accompanied by abrupt shifts in certainty or affect. As the system approaches a switching threshold, activity in competing populations may become comparable, generating a sense of indecision or conflict. Once the threshold is crossed and a particular attractor gains self-sustaining dominance, the competing state is rapidly suppressed, and associated doubts become less accessible to consciousness. This rapid suppression contributes to the phenomenology of having \u201cmade up one\u2019s mind,\u201d even though subthreshold traces of the alternative may persist in the circuitry and influence future decisions. When those traces are unexpectedly reactivated by new evidence or contextual change, the system can transition into the previously suppressed attractor, subjectively experienced as a sudden change of heart rather than as the outcome of continuous, low-level competition.<\/p>\n

Experimental approaches and future directions<\/h3>\n

Empirical progress on the two-state brain and bidirectional evidence accumulation hinges on experimental paradigms that can simultaneously track rapid neural dynamics, quantify behavioral state switches, and manipulate the underlying parameters that govern attractor stability. One foundational approach is to design tasks that reliably elicit bistable or near-bistable behavior, such as ambiguous percepts, multistable motion stimuli, or forced-choice decision tasks with finely titrated evidence strength. In these paradigms, researchers can measure the timing and frequency of perceptual reversals or changes of mind, using psychophysical methods and reaction-time distributions to infer the latent switching thresholds and integration windows. By systematically varying stimulus noise, context cues, and task instructions (for example, emphasizing speed versus accuracy), it becomes possible to map how external conditions reshape the effective attractor landscape that the neural system navigates.<\/p>\n

To connect these behavioral measures with the underlying circuitry, multi-modal neuroimaging and electrophysiological techniques are essential. High temporal resolution methods like EEG and MEG capture rapid oscillatory signatures associated with transitions between attractor states, allowing researchers to identify pre-switch markers such as increased variability, desynchronization, or characteristic phase resets. Concurrently, techniques with higher spatial resolution, including fMRI and two-photon calcium imaging in animal models, reveal which cortical and subcortical regions participate in each state and how their activation patterns reconfigure during transitions. Combining these modalities in the same subjects, or across complementary experiments, supports a richer characterization of the two-state regime, where distinct spatial patterns of activity correspond to alternative attractors and specific temporal motifs signal imminent switching.<\/p>\n

Single-unit and population recordings in animals provide a more granular view of how recurrent networks implement bidirectional evidence accumulation. In tasks where animals choose between two options based on noisy or conflicting cues, neurons in parietal, frontal, and basal ganglia circuits often show ramping activity consistent with evidence integration and abrupt changes in firing that track choice commitment. By aligning neural trajectories in state space to behavioral reports of perception or choice, experimenters can identify low-dimensional manifolds that correspond to each attractor and trace the pathways along which the system moves during state transitions. Perturbation techniques such as optogenetics, focal electrical stimulation, or chemogenetics then test the causality of these observations: transiently boosting or suppressing specific populations at key time points can bias the system toward one attractor, alter switching thresholds, or change the temporal profile of evidence accumulation.<\/p>\n

Closed-loop experimental designs are particularly powerful for probing the dynamics of the two-state brain. In these setups, neural signals or behavioral indicators of an impending switch are detected online, and targeted interventions are delivered in real time. For example, detecting a rise in EEG variability or a specific pattern of phase coupling that predicts a perceptual reversal can trigger a brief perturbation (such as transcranial magnetic stimulation or a change in stimulus statistics) designed to either promote or prevent the transition. By comparing trials in which the intervention succeeds or fails to alter the outcome, researchers can infer which neural signatures are merely correlates of switching and which play a causal role in crossing the threshold between attractors. This closed-loop methodology also makes it possible to explore how bidirectional evidence accumulation can be steered, not just passively observed.<\/p>\n

Noninvasive brain stimulation in humans provides a complementary route to manipulating attractor dynamics and testing theoretical predictions derived from bayesian inference and predictive processing. Transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), and transcranial alternating current stimulation (tACS) can modulate cortical excitability, oscillatory phase, and network coherence over specific regions involved in perceptual bistability and decision-making. By applying stimulation that enhances inhibition, increases noise, or biases oscillatory phase relationships, researchers can experimentally deepen or shallow attractor basins, thereby changing how resistant current states are to revision. For instance, enhancing frontal theta activity might increase the sensitivity of control networks to prediction errors, lowering switching thresholds and boosting cognitive flexibility, whereas boosting local gamma synchrony in sensory cortex could stabilize a particular percept against competing interpretations.<\/p>\n

Another crucial line of investigation focuses on reconstructing the implicit priors that drive attractor formation and maintenance. Behavioral paradigms that systematically manipulate prior probabilities, reward contingencies, and contextual frames allow estimation of the subjective prior distributions that participants use when interpreting ambiguous evidence. These priors can be inferred through model-based analyses that fit choice behavior and reaction times under different experimental conditions to computational frameworks grounded in bayesian inference. Once inferred, priors can be explicitly tested against neural measures: for example, deeper priors for a given hypothesis should be reflected in stronger baseline activation and more robust recurrent connectivity among neurons encoding that hypothesis before any evidence is presented. Longitudinal studies can then track how training, expectation manipulation, or feedback reshape both priors and the corresponding neural attractors over time.<\/p>\n

Advanced computational modeling plays a central role in integrating these diverse data streams and generating falsifiable predictions about two-state dynamics. Biophysical network models with realistic synaptic kinetics and cell types can simulate how excitation\u2013inhibition balance, adaptation mechanisms, and neuromodulatory effects give rise to bistable attractors and state transitions. Simpler dynamical systems models, such as coupled drift\u2013diffusion processes or winner-take-all networks, can be used to fit behavioral and neural data directly, linking observable quantities like switch rates, decision bounds, and oscillatory power to underlying parameters like attractor depth and noise levels. Critically, these models must incorporate bidirectional influences: top-down signals that alter priors and context, and bottom-up signals that encode prediction errors. Fitting such models to experimental data provides estimates of how strongly predictive processing and priors shape the effective energy landscape in which the two-state brain operates.<\/p>\n

To validate and refine these models, researchers increasingly deploy multivariate decoding and representational similarity analysis on neural recordings. Decoders can be trained to distinguish which attractor state the system is currently occupying based on patterns of activity across many channels. By tracking decoder output over time, one can identify not only discrete switches between states but also transient mixtures, pre-switch drifts, and partial reactivations of previously suppressed attractors. Representational similarity analyses reveal how the geometry of neural state space reorganizes during learning or under different task demands: for example, whether attractors corresponding to two conflicting interpretations move farther apart as a subject acquires expertise, or whether they become more entangled under conditions of high uncertainty. These methods bridge the gap between low-level firing patterns and higher-level theoretical constructs such as state stability and transition thresholds.<\/p>\n

Looking ahead, one major frontier lies in extending two-state and bidirectional accumulation frameworks to richer, multistate and hierarchical settings that better approximate everyday cognition. Real-world decisions rarely involve only two discrete alternatives; instead, networks must negotiate among many competing hypotheses at multiple levels of abstraction, from low-level sensory features to high-level goals and narratives. Experimental paradigms that involve complex, naturalistic tasks\u2014such as navigating virtual environments, understanding stories, or engaging in social interactions\u2014will be essential for probing how local two-state competitions are embedded within larger graphs of attractors. Here, tools like virtual reality, interactive gaming tasks, and ecologically valid decision-making scenarios can be combined with mobile neuroimaging technologies to observe how state transitions unfold in more lifelike conditions.<\/p>\n

Another promising direction involves investigating how two-state dynamics intersect with different modes of consciousness, including sleep, anesthesia, and altered states induced by psychoactive substances. Changes in global neuromodulatory tone and large-scale network connectivity in these conditions provide natural perturbations to the attractor landscape. For example, anesthesia and deep sleep may flatten or fragment attractor basins, reducing the capacity for stable global configurations and thus limiting conscious access, whereas certain psychedelics may transiently lower switching thresholds and increase the fluidity of transitions between attractors. Systematic comparisons of neural dynamics across these conditions, especially using techniques that quantify integration and segregation in large-scale networks, can shed light on how the capacity to occupy and transition between stable states relates to the presence and richness of conscious experience.<\/p>\n

Clinical research offers a further avenue for testing and refining theories of the two-state brain. Disorders characterized by excessive rigidity, such as obsessive\u2013compulsive disorder or certain forms of depression, and those characterized by instability, such as bipolar disorder or attention-deficit\/hyperactivity disorder, can be viewed through the lens of altered attractor depth and maladaptive switching thresholds. Experimental protocols that combine behavioral tasks, neuroimaging, and pharmacological or neuromodulatory interventions can assess how treatments change the structure of the attractor landscape. For instance, response to antidepressants or cognitive-behavioral therapy might be accompanied by measurable increases in the frequency of adaptive state transitions, reductions in the dominance of negative affective attractors, or enhanced capacity to shift between task sets. These findings could, in turn, inform personalized treatment strategies that explicitly aim to normalize two-state dynamics.<\/p>\n

Emerging technologies in large-scale neural recording and intervention promise even more fine-grained tests of the bidirectional evidence accumulation framework. High-density electrode arrays, optical imaging of genetically defined cell types, and closed-loop optogenetic control in animal models make it possible to monitor and manipulate thousands of neurons across multiple brain regions in real time. With these tools, one can directly observe how local state transitions propagate across a distributed network, how top-down and bottom-up signals interact during ambiguous decisions, and how small perturbations at one node cascade into global reorganizations of activity. By systematically varying the timing, location, and strength of perturbations, researchers can map causal pathways of influence and delineate which circuits are critical for maintaining state stability, which are responsible for initiating transitions, and which mediate bidirectional communication between levels of the hierarchy.<\/p>\n

There is growing interest in leveraging insights from the two-state brain to inspire new algorithms and architectures in artificial intelligence. Machine learning models that incorporate attractor dynamics, recurrent communication between levels, and explicit representations of priors are well positioned to implement more human-like forms of flexible inference. Experimental collaborations between neuroscience and AI can explore how concepts like attractor depth, switching thresholds, and bidirectional evidence accumulation translate into improved robustness, sample efficiency, and interpretability in artificial systems. In return, synthetic models can generate precise, testable hypotheses about which patterns of activity or connectivity should be observable in biological brains under specific conditions, guiding future empirical work. As experimental approaches become more sophisticated and theoretical frameworks more tightly coupled to data, these cross-disciplinary efforts are likely to sharpen our understanding of how the two-state brain negotiates the flow of information between past evidence, present context, and anticipated futures.<\/p>\n","protected":false},"excerpt":{"rendered":"

The notion of a two-state brain captures the idea that large-scale neural activity often settles…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","_lmt_disableupdate":"","_lmt_disable":"","footnotes":""},"categories":[1],"tags":[333,371,1617,1615,1613,1743],"class_list":["post-3163","post","type-post","status-publish","format-standard","hentry","category-uncategorized","tag-bayesian-inference","tag-consciousness","tag-predictive-processing","tag-priors","tag-retrocausality","tag-two-state-vector"],"yoast_head":"\nThe two-state brain and bidirectional evidence - Beyond the Impact<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/beyondtheimpact.net\/?p=3163\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"The two-state brain and bidirectional evidence - Beyond the Impact\" \/>\n<meta property=\"og:description\" content=\"The notion of a two-state brain captures the idea that large-scale neural activity often settles…\" \/>\n<meta property=\"og:url\" content=\"https:\/\/beyondtheimpact.net\/?p=3163\" \/>\n<meta property=\"og:site_name\" content=\"Beyond the Impact\" \/>\n<meta property=\"article:published_time\" content=\"2025-12-21T00:01:32+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"40 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163#article\",\"isPartOf\":{\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163\"},\"author\":{\"name\":\"admin\",\"@id\":\"https:\/\/beyondtheimpact.net\/#\/schema\/person\/a5cf96dc27c4690dbf266a6cae4ee9aa\"},\"headline\":\"The two-state brain and bidirectional evidence\",\"datePublished\":\"2025-12-21T00:01:32+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163\"},\"wordCount\":8081,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/beyondtheimpact.net\/#organization\"},\"keywords\":[\"Bayesian inference\",\"consciousness\",\"predictive processing\",\"priors\",\"retrocausality\",\"two-state vector\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/beyondtheimpact.net\/?p=3163#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163\",\"url\":\"https:\/\/beyondtheimpact.net\/?p=3163\",\"name\":\"The two-state brain and bidirectional evidence - Beyond the Impact\",\"isPartOf\":{\"@id\":\"https:\/\/beyondtheimpact.net\/#website\"},\"datePublished\":\"2025-12-21T00:01:32+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/beyondtheimpact.net\/?p=3163\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/beyondtheimpact.net\/?p=3163#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/beyondtheimpact.net\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"The two-state brain and bidirectional evidence\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/beyondtheimpact.net\/#website\",\"url\":\"https:\/\/beyondtheimpact.net\/\",\"name\":\"BeyondTheImpact\",\"description\":\"Concussion, FND and Neuroscience\",\"publisher\":{\"@id\":\"https:\/\/beyondtheimpact.net\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/beyondtheimpact.net\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/beyondtheimpact.net\/#organization\",\"name\":\"Beyond the Impact\",\"url\":\"https:\/\/beyondtheimpact.net\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/beyondtheimpact.net\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/beyondtheimpact.net\/wp-content\/uploads\/2025\/04\/955D378D-9439-4958-AA9D-866B66877DCB-1.png\",\"contentUrl\":\"https:\/\/beyondtheimpact.net\/wp-content\/uploads\/2025\/04\/955D378D-9439-4958-AA9D-866B66877DCB-1.png\",\"width\":1024,\"height\":1024,\"caption\":\"Beyond the Impact\"},\"image\":{\"@id\":\"https:\/\/beyondtheimpact.net\/#\/schema\/logo\/image\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\/\/beyondtheimpact.net\/#\/schema\/person\/a5cf96dc27c4690dbf266a6cae4ee9aa\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/beyondtheimpact.net\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/59867129c03db343d7fdc6272ec5e0a85250cd376a4e7153307728ae82a1b108?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/59867129c03db343d7fdc6272ec5e0a85250cd376a4e7153307728ae82a1b108?s=96&d=mm&r=g\",\"caption\":\"admin\"},\"sameAs\":[\"https:\/\/beyondtheimpact.net\"],\"url\":\"https:\/\/beyondtheimpact.net\/?author=1\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"The two-state brain and bidirectional evidence - Beyond the Impact","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/beyondtheimpact.net\/?p=3163","og_locale":"en_US","og_type":"article","og_title":"The two-state brain and bidirectional evidence - Beyond the Impact","og_description":"The notion of a two-state brain captures the idea that large-scale neural activity often settles…","og_url":"https:\/\/beyondtheimpact.net\/?p=3163","og_site_name":"Beyond the Impact","article_published_time":"2025-12-21T00:01:32+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"Written by":"admin","Est. reading time":"40 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/beyondtheimpact.net\/?p=3163#article","isPartOf":{"@id":"https:\/\/beyondtheimpact.net\/?p=3163"},"author":{"name":"admin","@id":"https:\/\/beyondtheimpact.net\/#\/schema\/person\/a5cf96dc27c4690dbf266a6cae4ee9aa"},"headline":"The two-state brain and bidirectional evidence","datePublished":"2025-12-21T00:01:32+00:00","mainEntityOfPage":{"@id":"https:\/\/beyondtheimpact.net\/?p=3163"},"wordCount":8081,"commentCount":0,"publisher":{"@id":"https:\/\/beyondtheimpact.net\/#organization"},"keywords":["Bayesian inference","consciousness","predictive processing","priors","retrocausality","two-state vector"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/beyondtheimpact.net\/?p=3163#respond"]}]},{"@type":"WebPage","@id":"https:\/\/beyondtheimpact.net\/?p=3163","url":"https:\/\/beyondtheimpact.net\/?p=3163","name":"The two-state brain and bidirectional evidence - Beyond the Impact","isPartOf":{"@id":"https:\/\/beyondtheimpact.net\/#website"},"datePublished":"2025-12-21T00:01:32+00:00","breadcrumb":{"@id":"https:\/\/beyondtheimpact.net\/?p=3163#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/beyondtheimpact.net\/?p=3163"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/beyondtheimpact.net\/?p=3163#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/beyondtheimpact.net\/"},{"@type":"ListItem","position":2,"name":"The two-state brain and bidirectional evidence"}]},{"@type":"WebSite","@id":"https:\/\/beyondtheimpact.net\/#website","url":"https:\/\/beyondtheimpact.net\/","name":"BeyondTheImpact","description":"Concussion, FND and Neuroscience","publisher":{"@id":"https:\/\/beyondtheimpact.net\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/beyondtheimpact.net\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/beyondtheimpact.net\/#organization","name":"Beyond the Impact","url":"https:\/\/beyondtheimpact.net\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/beyondtheimpact.net\/#\/schema\/logo\/image\/","url":"https:\/\/beyondtheimpact.net\/wp-content\/uploads\/2025\/04\/955D378D-9439-4958-AA9D-866B66877DCB-1.png","contentUrl":"https:\/\/beyondtheimpact.net\/wp-content\/uploads\/2025\/04\/955D378D-9439-4958-AA9D-866B66877DCB-1.png","width":1024,"height":1024,"caption":"Beyond the Impact"},"image":{"@id":"https:\/\/beyondtheimpact.net\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/beyondtheimpact.net\/#\/schema\/person\/a5cf96dc27c4690dbf266a6cae4ee9aa","name":"admin","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/beyondtheimpact.net\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/59867129c03db343d7fdc6272ec5e0a85250cd376a4e7153307728ae82a1b108?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/59867129c03db343d7fdc6272ec5e0a85250cd376a4e7153307728ae82a1b108?s=96&d=mm&r=g","caption":"admin"},"sameAs":["https:\/\/beyondtheimpact.net"],"url":"https:\/\/beyondtheimpact.net\/?author=1"}]}},"_links":{"self":[{"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=\/wp\/v2\/posts\/3163","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3163"}],"version-history":[{"count":0,"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=\/wp\/v2\/posts\/3163\/revisions"}],"wp:attachment":[{"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3163"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3163"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/beyondtheimpact.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3163"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}