Goal-oriented behaviors are the seminal component of decision-making, vital for survival and exhibited in a time-dependent manner, reflecting the brain’s chrono-properties (Honma, 2020), which have recently gained increased attention (Gupta et al., 2020). Cells sensitive to time differences were identified in several brain regions (MacDonald et al., 2011), and much interest has been focused on the temporal aspects underlying overt behavior. Together, this suggests a complex view beyond the accepted measure of behavioral reaction time and the simple discrimination between rapid and slow decision-making.

Control Systems Theory (CST), initially described mathematically by Maxwell in the 19th century, underwent major developments since then (e.g., Lyshevski, 2001; Schiff, 2011). This theory implies dynamic control of signals by internal controllers implemented in the system, to attain a required level of a signal (set point) compared to an actual level of that signal. Following that comparison, the controller produces a corrected output. Consistent feedback comparisons are hence an essential part of CST and are applicable for the behavioral outcomes, as outputs of the CNS (Bennett, 1993; Goldstein Ferber et al., 2022). To avoid conflicts between set points in a distributed (multi-level) control system, the logic gates in each controller need to function in concert with each other. Thus, each control system functions according to its implemented “timers.”

We suggest that the transcriptional and metabolic constructs underlying goal-oriented behaviors operate as parallel sequences of a control system. We adapt the term “parallel sequences” from the field of engineering (Williams et al., 2017) to describe this multi-level regulation.

Our application of CST follows the conceptualization of Neuro-constructivism, suggesting that goal-directed behavior derives from mental representations resulting from experience-dependent development of biological constructs (Mareschal et al., 2007a,b; Sansavini et al., 2021; Farran and Scerif, 2022). Neuro-constructivism outlines how the brain subjects cognition events to aftermath processing to reach its ultimate function, goal-oriented behavior (Sirois et al., 2008).

We incorporate Neuro-constructivism concepts discriminating between regulated and non-regulated neural bases for goal-oriented behavior, founded on CST Modeling, using Boolean Algebra (Ferber, 2008). Considering that every neuronal circuit involves numerous logic gates (Ferber, 2008; von Neumann, 2017), we argue that parallel sequences, beyond simple paired-pulse facilitation, enable logic gates crosstalk at a multi-level basis reflecting a spatially encoded chrono-property of the healthy brain. This temporal consideration spans the atomic, molecular and cellular levels, covering local populations of neurons and their supporting cell types within and between the brain’s circuitry and regions. Furthermore, the conceptual parallel sequences suggest simultaneous matching of the new code with the old one, without losing previously coded information (Ferber, 2008). Based on thermodynamic, electrophysiological and neuro-molecular principles, our concept argues that within the brain, multi-level metabolic “timers” determine the time-to-action in parallel sequences and within reversible thermodynamic states.

This crosstalk between the multi-level metabolic “timers” is suggested here as parallel sequences with propagation delays of signaling at each level, in the beginning and the end of each sequence (Williams et al., 2017). The parallel sequences are apparent within each level and between levels. The efficacy of timescales is determined by the regulation of the output of these sequences, as an XOR gate. This is a function of availability of free energy at a given moment for a given internal or environmental stimulus, which in turn is limited by the given level of entropy. According to the Boltzmann Theorem, micro-states are accumulated into a macro-state of available thermodynamic free energy for a given action at a given time. Furthermore, as implied by this theorem, a macro-state exhausts itself, with a given behavioral reaction, which, however, may be reversible, when new micro-states accumulate to a new macro state, even for repeated action. In this sense, the occurrence of entropy and the exhaustion of a macro-state and a given gathered free energy for an action implies the beginning of the next macro-state. Although this claim of reversible macro-states may be understood as serial, we argue that they are computed by the different brain levels in parallel sequences, meaning numerous macro-states for emotional, cognitive and behavioral outputs at the same time.

Adding metabolic elements to CST enables viewing the brain as a thermodynamic system (Collell and Fauquet, 2015), involving a free energy-entropy matrix. Thermodynamic free energy reflects the work (useful energy) which a system can perform at constant temperature, i.e., the energy quantity available for transition or change. The corresponding temporal component is defined by the thermodynamic free energy quantities available for transition or change (Hill, 1977; Friston et al., 2006; Friston and Klaas, 2007).

According to the second law of thermodynamics, within isolated systems, entropy tends to increase over time. As the brain is not an isolated system, we suggest that the environment may alter the entropic speed in a neural network and thus change the preprogrammed availability of free energy for a behavioral action to occur (Schrodinger, 1951).

We further adopt Boltzmann’s theorem, which defines systems in thermodynamic equilibrium as capable of creating a corresponding microstate of available energy while keeping microscopic states available to the system, dependent on macroscopic limitations; meaning the system’s total level of entropy. Entropy is the probability of the tendency to proceed toward the most probable macro-state over time (Darwin, 1909; Jeffery and Rovelli, 2020). Others stated that once a system has reached its thermodynamic equilibrium, its entropy is maximized, and no further change is possible (Andelman, 1995; Villani, 2008; Brown et al., 2009; Bertolami and Gomes, 2020). In contradistinction, we posit that no true end exists for any given macro-state in a live brain throughout the life span of its neurons and supporting cell types, as each macro-state initiates the next cycle toward its following thermodynamic equilibrium. Several series of such cycles may appear in concert, such that establishing a macro-state at a certain time requires maximal total available metabolic energy for a given brain at a given moment while controlling, in parallel, the maximum number of cerebral micro-state sequences. Since there are many methods to measure entropy, we refer here to multiscale entropy (MSE), that takes times and rhythms into the account of entropy levels (Gao et al., 2015; Jaworska et al., 2018).

We hypothesize that logic gates represent molecular and metabolic constructs and that XOR gates represent regulated goal-oriented behavior (Ferber, 2008). Correspondingly, XOR gates (including XNOR gates) reflect local neural networks’ excitation-inhibition balance in vivo (Goldental et al., 2014). Our current model (Figure 1) argues that a goal-oriented behavior depends on a temporal range of optimal signal variability based on metabolic processes, at levels spanning from atoms though molecular, cellular, network populations and inter-regional signaling, all of which operate as parallel sequences (Williams et al., 2017).

Our model involves serial parallel controllers (1 to N at each level) communicating with each other, at scales ranging from atoms through macromolecules, cells, local and large networks up to the communication between regions, all in parallel sequences (at the temporal level) from micro- to macro-thermodynamic states. To determine the final output stimulated by environmental and internal requirements (input), each controller has its own set-point for comparison processes. Set-points are intrinsic changeable capacities when the system’s environmental exposure is altered or triggered by developmental trajectories and progression in one’s age. Each unit controller consists of millions of logic gates (micro-states) whose truth tables undergo a minimization process (see Table 1, following Taub, 1982). Pruning or selection of synapses toward reaching a final XOR gate condition would hence activate a healthy and adaptive behavioral reaction (Figure 1).

While the elements within the parallel sequences are standardly ordered when operating within the same timeframe, that timeframe depends on the range of the required signal variability (Verschure et al., 2014; Sendhilnathan et al., 2021) for a given goal-oriented behavior. To standardize the parallel sequences within a timeframe (the overall healthy range of signal variability) (Northoff, 2016), such standards may involve propagation delays at the beginning or end of the sequence, depending on its type (Williams et al., 2017). The healthy clock of each goal-oriented behavior hence comprises a fixed number of parallel sequences. Those are activated by inter-related environmental and internal stimuli (input), and processed at a given, age-appropriate moment, requiring an adaptive reaction (output).

The variables in our model determine the Truth Table for its logic gates (Table 1). Hence, pruning involves normal minimization of the original numerous logic gates (i.e., metabolic constructs) needed for the brain’s network to release a goal-directed behavior. The temporal component of these parallel sequences is determined by the ability of the macro-state to achieve a final XOR gate (regulated) condition for a healthy, goal-directed behavior. Its duration depends on the type of the environmental stimulus that requires a goal-directed action and the available inhibitory and excitatory synapses (actual value) for its adaptive activation, from all needed synapses for this action (required value).

Describing the temporal range initiated by external stimulation (input) up to the resulting behavior (output) spans the ascending (sensory) tracts, where levels ranging from atoms to regions (microstates) are triggered by environmental and internal stimuli; and the descending (motor) tracts, where goal-directed behavior reflects the entropy-free energy matrix (available macro-state at a given moment, age-appropriate).

This model implements metabolic constructs (1) at the atom level (2) ion channels at the cellular level (3) signal variability of proteins at the local network level and (4) cortical gradients for the inter-regional level.

Our model differs from models presented earlier in showing that XOR gates represent an overarching excitation-inhibition balance across multiple levels, which comprises micro- to macro-thermodynamic reversible states of the brain. Supporting this view, XOR gates have been shown to represent local neural networks’ excitation-inhibition in neural network simulations (Vogels and Abbott, 2005) and in vivo (Goldental et al., 2014), in accordance with our propositions.

Our model may be adapted to either narrower or broader prisms\spectra of neural and brain functioning by defining distinct numbers of levels. This may involve aligned Truth Tables and their minimization for XOR gate output and may generate further hypotheses. Our overarching hypothesis relates to regulating the excitation-inhibition balance from atoms through molecular, cellular network and inter-regional levels. We investigate the interim pathways from the genetic quasi-deterministic levels of time-to-action to experienced-modified signal variability and its optimal range (Goldstein Ferber et al., 2022). This could be examined by in-vivo tests of the healthy signal variability range and its XOR gate functionality. Such studies may open novel opportunities for more accurate and reliable diagnoses and improved prevention and treatment of psychopathologies.

The abundant mental disorders in humans are presumably rooted in an evolutionary environmental-biology mismatch, and may predictably interrupt XOR processes. Nevertheless, pinpointing the biological mechanisms that regulate the corresponding behaviors and their mental consequences is challenging. Recently, a novel class of non-coding RNA molecules, called transfer RNA fragments (tRFs) was identified as playing critical roles in timing brain activities. However, how are these factors dysregulated in disease, what are the molecular mechanisms through which tRFs act in response to stress and to what extent are these mechanisms evolutionarily conserved remains unknown.

Gene regulation is a fundamental process that dictates which genes of a cell are actively expressing functional gene products, including RNA and proteins, at any given time (Soreq, 2015). Even though each brain cell contains the same genetic code, different genes are expressed in each cell context-dependently, through multi-step regulatory processes. These events define the identity and function of a cell and can be subject to change, depending on the internal and the external environments experienced during the life course and mental state of an individual.

Non-coding RNAs (ncRNAs) are biological molecules regulating gene expression and cellular function. Historically, they have been thought to be ‘junk’ RNA, originating from errant transcription, or breaking down of no longer needed RNA molecules. However, some ncRNA classes were shown to regulate the expression of genes in a human disease-dependent manner. Recently re-discovered among ncRNAs are transfer rNA fragments (tRFs), which originate from the targeted fragmentation of tRNAs that carry amino acids to the ribosome to enable protein synthesis. Recent findings showed that tRFs are not random fragmentation products of cellular metabolism, but rather a novel, understudied class gene expression regulators, whose biogenesis is controlled by precise and conserved site-specific cutting mechanisms (e.g., Winek et al., 2020, 2021) in human health and disease.

While tRFs have recently been shown to have functional features similar to those of microRNAs (miRs), it is still unknown if they regulate behavior. Several tRFs contribute to embryonic development and neuronal function, with their abundance changing with age (Karaiskos et al., 2015; Cao et al., 2021; Paldor et al., 2022), including tRFs originating from the tRNA LysTTT. While their complex impact on brain-to-body functioning is largely unresolved, small regulatory RNAs are rapidly acquiring wide recognition as global controllers of regulatory processes at large and may emerge as controllers of the above-described brain clocks. Thus, tRFs may emerge as the initiators (the “onset”) of time-dependent parallel sequences ranging from atom levels through molecular, cellular, local network and interregional signaling. Supporting this notion, tRFs are sensitive to environmental stimuli and able to trigger the available metabolic processes (e.g., Winek et al., 2020) termed here as thermodynamic macro-states for an action to be performed. A further prediction, based on the findings on the regulatory role of tRFs (e.g., Dubnov and Soreq, 2023), is that tRFs will facilitate the regulatory output of in vivo XOR gates within the parallel sequences at the different levels determining time-to-action functionality.

One possible design to test our model could be to concurrently sample event-related H1-magnetic resonance spectroscopy (MRS) and electroencephalography (EEG) as in Lally et al. (2014), across different environmental stimuli and at different ages. The MRS can provide measures on the levels from atoms through molecules to cells and the EEG can detect local neural networks and inter-regional connectivity.

We view the brain clocks of time-to-action as a control system dependent on the rhythms of regulated metabolic constructs for its healthy and timely signaling. The entire range from atomic, molecular, cellular, local circuitry and inter-regional network levels operates in parallel, computing concomitantly sequential signals from the lowest atomic level up to the highest level of inter-regional neurite functionality. This sequential simultaneous signaling represents combined parallel processing for the final measure of time-to-action. We characterize healthy time-to-action by its excitation-inhibition balance at each level being within an optimal range of the individual’s chrono-properties. Clock genes are designed for enabling initiation of these parallel processes, calculations of free energy and derived optimal time-to-action.

The brain is an open thermodynamic system and its levels of entropy increase or decrease by interacting with the environment, which may be under-stimulating, stressful or regulatory, thus determining the sum of available synapses. Additional factors including age, preprogrammed developmentally appropriate steps (Ferber et al., 2022) and previous encounters affect pruning and experienced-shaped behaviors (Sirois et al., 2008), and determine available synapse numbers compared to the required number for a regulated, age-appropriate time-to-action at a given moment.

In sum, the clocks operate at atoms, cells, local neural populations and inter-regional levels. The healthy clocks of time-to-action regulate XOR logic gates communicating between micro- and macro-states. Imbalanced excitation-inhibition, represented by deviations from XOR gates to other types of logic gates, shows suppressed healthy pruning. Finally, experience-shaped neural structures and their macro-state in a given environmental exposure determine the underlying parallel sequences of clocks required for an action.

The original contributions presented in this study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

SGF was the leading author. All authors contributed substantially to the different versions of the manuscript and to its final version.