Brains May Use Heat Flow to Process Information, Study Suggests

Onur Pusuluk and colleagues at Kadir Has University present a multiscale theoretical framework centred on the ‘thermocoherent effect’, proposing a fundamental link between heat flow and information flow through shared coherence. The framework moves beyond classifying correlations, focusing instead on their dynamic accessibility and potential as a usable physical resource within neural matter. By suggesting that relational structures, including entanglement and classical correlations, can bias fundamental transport processes, the researchers offer a falsifiable framework for understanding how microscopic interactions may contribute to larger-scale neural organisation and cognitive dynamics. The investigation concerns the physical mechanisms underpinning information flow within living neural tissue, a long-standing challenge in understanding cognition. Thermodynamic constraints shaping relational resources in neural tissue Electrical, chemical, ionic, and thermal transport processes in neural matter can generate or transduce partially hidden relational resources under suitable microscopic conditions. The mutual coupling of these resources progressively builds larger-scale thermocoherent organisation across spatial or spatiotemporal partitions in neural tissue. Ion-channel interfaces, hydrogen-bonded proton networks, aromatic π-electron architectures, and phosphate-rich motifs plausibly serve as substrate classes where such resources may arise, become transiently accessible under environmental coupling, and leave coarse-grained signatures in neural dynamics. The resulting picture is neither a claim of macroscopic quantum cognition nor a reduction of cognition to Abstract coding, but a falsifiable framework where microscopic relational resources can bias transport, relaxation, signalling, and cross-scale neural coordination. Information flow plays a central role in contemporary accounts of cognition, but the physical entity or process it corresponds to remain unclear. Many descriptions treat information as a largely Abstract, phenomenological, or coding-level notion, or as fully specified by the nonequilibrium organisation of the material substrates that support it (see Fig0.1). These viewpoints, when considered in isolation, can lead to complementary limitations: either detaching information from the physical transport processes that implement and constrain it, or identifying it too directly with the local physical currents that accompany it. Recent developments in quantum information thermodynamics suggest a more subtle possibility: information is a physical resource that need not be reducible to local subsystem properties and can, under suitable conditions, be distributed or shared through relational structure, including correlations. This possibility opens the door to grounding information flow physically without reducing it to subsystem-local transport alone. It also brings into sharper focus a reciprocal question that remains conceptually underdeveloped in many cognitive and neuroscientific accounts: how information-bearing organisation, once physically instantiated, can constrain, redirect, or modulate subsequent material dynamics. The thermocoherent effect provides a particularly clear example of this phenomenon. In that setting, heat flow is reciprocally coupled to a delocalized information flow carried by quantum coherence shared between degenerate energy levels, so that the latter is not localised within either subsystem even though it can directly influence energy transport. This feature distinguishes the thermocoherent effect from more generic coherence-assisted thermodynamic phenomena: its key novelty is not merely that coherence modifies heat flow, but that an explicitly delocalized information flow emerges alongside it, remains reciprocally coupled to the heat current, and thereby mediates the redistribution of relational structure across subsystems, opening a distinct route by which hidden organisation can feed back onto energy transport. More recent resource-theoretic analyses, including correlation-enabled Mpemba-type relaxation scenarios, broaden this lesson further. In such cases, thermodynamically active hidden structure need not be restricted to coherence-based resources alone. Under suitable interaction geometry and spectral accessibility, operationally relevant structure may also be carried by classical correlations encoded in the state of a single composite system, even when the local marginals remain thermodynamically indistinguishable. Once such hidden relational structure becomes dynamically accessible, it can alter relaxation ordering and effective cooling behaviour. This claim requires a precise use of the term “classical.” Here, “classical” does not refer to ensemble-level statistical dependence, trial-averaged covariance, or a merely epistemic lack of knowledge. Rather, it denotes physically instantiated relational structure in an individual composite system, as represented by its density operator. Nonclassical correlations, including quantum entanglement and quantum discord, are understood in the same state-level sense. The present framework therefore concerns relational resources internal to single composite systems, not coarse-grained statistical correlations across repeated trials. This distinction matters operationally, because systems with indistinguishable local marginals may still differ substantially in hidden relational structure that becomes dynamically relevant only under suitable couplings. The scope of this perspective extends beyond single-time state-level descriptions. It need not be confined to relational structure across spatial partitions at a single instant. Depending on the operational setting, relevant hidden structure may also appear in temporally ordered or multi-time relational form that is inaccessible even to instantaneous joint-system descriptions yet still constrains later transport, response, or route selection. In history-dependent settings, such process-level relational structure may also provide a natural language for discussing memory-like behaviour and certain forms of non-Markovianity. With this background, the aim of the present work is to extend this substrate-agnostic thermodynamic perspective toward biological and neural matter without introducing a new neuroscientific primitive or presupposing macroscopic quantum cognition. The work asks whether coexisting electrical, chemical, ionic, and thermal transport processes in living systems, and in particular in neural matter, can support forms of relational organisation whose redistribution, dynamical accessibility, or temporal persistence become operationally relevant in ways analogous to the thermocoherent and correlation-enabled effects discussed above. David Abbott and Matthew Clarkson first introduce the minimal formal distinctions used throughout the rest of this work before turning to operational archetypes and candidate substrate classes. At this stage, the aim is simply to identify how relational structure can be encoded in composite descriptions, how it may remain hidden from subsystem-reduced views, and under what conditions it can become operationally accessible. Consider two subsystems A and B with basis states {|0⟩A, |1⟩A} and {|0⟩B, |1⟩B}. Their local states are diagonal in these bases, ρA= p|0⟩A⟨0| + (1 −p) |1⟩A⟨1|, ρB= q|0⟩B⟨0| + (1 −q) |1⟩B⟨1|, so that p and q are the populations of the local |0⟩ states, while 1 −p and 1 −q are the populations of the local |1⟩ states. For pedagogical concreteness, A and B may be interpreted as two localised physical modes relevant to a given substrate class, such as proton-occupancy sites along a hydrogen-bond network or ion-occupancy sites within a channel selectivity filter. In realistic neural settings, the local basis states |0⟩ and |1⟩ may represent absence/presence, unoccupied/occupied, or spin-down/spin-up configurations depending on the context. The formal structure developed below is independent of the particular interpretation. The product state ρuc AB= ρA⊗ρB is called uncorrelated and reads ρuc AB= © « pq 0 0 0 0 p(1 −q) 0 0 0 0 (1 −p)q 0 0 0 0 (1 −p)(1 −q) a®®® ¬ in the computational product basis {|00⟩, |01⟩, |10⟩, |11⟩}. This state contains no relational structure beyond the local marginals. A simple classically correlated state with the same local marginals can then be written as ρcc AB(χ) = ρuc AB+ Δχ, with Δχ= © « χ 0 0 0 0 −χ 0 0 0 0 −χ 0 0 0 χ a®®® ¬ , where χ is restricted by positivity. The reduced states of ρcc AB(χ) are still ρA and ρB, but the joint probabilities are no longer products of the marginals. Consequently, the relational structure of the composite system has changed even though the local descriptions have not. For every nonzero allowed value of χ, the state ρcc AB(χ) is non-product and carries genuine classical correlation. Information flow is central to contemporary accounts of cognition, yet its physical basis in living neural matter remains poorly specified. A multiscale resource-theoretical framework is developed, motivated by the thermocoherent effect, where heat flow is reciprocally coupled to a delocalized information flow carried by shared coherence and not reducible to local subsystem variables. Extending this line of work in light of recent results on correlation-enabled Mpemba-type thermal relaxation, it is argued that the operational relevance of correlations depends less on their taxonomy than on their dynamical accessibility under the underlying interaction geometry. Relational structure encoded in the state of a single composite system, including quantum entanglement, quantum discord, and classical correlations, may therefore act as a usable physical resource that remains hidden from local subsystem descriptions. Electrical, chemical, ionic, and thermal transport processes in neural matter may, under suitable microscopic conditions, generate or transduce partially hidden relational resources whose mutual coupling can progressively build larger-scale thermocoherent organisation across spatial or spatiotemporal partitions in neural tissue. Ion-channel interfaces, hydrogen-bonded proton networks, aromatic π-electron architectures, and phosphate-rich motifs emerge as plausible substrate classes in which such resources may arise, become transiently accessible under environmental coupling, and leave coarse-grained signatures in neural dynamics. For every nonzero allowed value of λ, this state carries nonzero shared coherence and, in this construction, nonzero quantum discord, making it a minimal representative of a nonclassical but non-entangled relational resource. An entangled counterpart may then be constructed by adding shared coherence to the classically correlated background, ρqe AB(μ) = ρcc AB(χmax) + Δμ, with Δμ= © « 0 0 0 μ 0 0 0 0 0 0 0 0 μ∗ 0 0 0 a®®® ¬ , where μ is restricted by positivity. In this case, the coherence resides in the |00⟩↔|11⟩ sector. For every nonzero value of μ, the resulting state exhibits entanglement while maintaining the same local marginals. Relational structure encoded within a composite system, encompassing quantum entanglement, quantum discord, and classical correlations, may function as a usable physical resource remaining hidden from descriptions of local subsystems. The three parameters χ, λ, and μ each leave the subsystem marginals unchanged while activating a distinct relational sector of the global state, namely classical correlation, discordant separability, and entanglement, respectively. These instances demonstrate the minimal distinction employed throughout this framework. Two composite systems may share identical local marginals while differing in the relational structure encoded in their joint density operators. Whether such differences remain dynamically silent or become operationally relevant depends on the interaction geometry and spectral accessibility of the coupling. When an interaction redistributes, transfers, reshapes, or converts accessible relational structure across different partitions of a larger composite state, what changes is not merely a local observable but the pattern of joint organisation itself. In this sense, one may speak of a delocalized information flow. In correlation-dominated settings, this may also appear as a redistribution of correlations across subsystem partitions, but the notion is broader than correlation flow alone.

Relational Structures Bias Transport and Unlock Neural Resources Relational structure, previously limited to assessing local subsystem properties, can now bias transport processes by a factor of χ, λ, and μ, enabling access to hidden resources within single composite systems. This threshold, unattainable in prior models focused solely on local variables, allows for the investigation of how microscopic interactions influence larger-scale neural organisation and cognitive dynamics. The framework extends the understanding of the thermocoherent effect, where heat flow is coupled to delocalized information, to encompass classical correlations alongside quantum entanglement and discord. By proposing that these relational resources act as usable physical entities, the work offers a falsifiable model for cognition, moving beyond abstract coding and macroscopic quantum interpretations. David Abbott and Matthew Clarkson identified specific molecular architectures within neural tissue potentially capable of hosting these relational resources, including ion-channel interfaces, hydrogen-bonded proton networks, and phosphate-rich motifs. These structures could generate or transduce partially hidden relational resources during electrical, chemical, ionic, and thermal transport processes, building larger-scale thermocoherent organisation across neural tissue. Furthermore, the framework reinterprets electromagnetic field-based binding proposals not as primary carriers of cognition, but as emergent coordination layers shaped by underlying thermodynamic constraints. The authors acknowledge a significant tension with existing theories positing electromagnetic fields as primary carriers of consciousness, such a view requiring a fundamental shift in perspective. The research demonstrated that relational structure within a single composite system can bias transport processes by factors of χ, λ, and μ, enabling access to previously hidden resources. This finding suggests information flow in neural tissue is not limited to local variables, but can be influenced by interactions between electrical, chemical, ionic, and thermal processes. Researchers identified ion-channel interfaces, hydrogen-bonded proton networks, and phosphate-rich motifs as potential substrates where these resources may arise. The authors propose further investigation is needed to fully understand how these relational resources contribute to larger-scale thermocoherent organisation in neural tissue. 👉 More information 🗞 The physical basis of information flow in neural matter: a thermocoherent perspective on cognitive dynamics 🧠 ArXiv: https://arxiv.org/abs/2604.04069 Tags: