The groove and the circuit

A practitioner who has sat for a decade still reaches for the phone the instant a meeting ends. He knows the reach is empty, has traced it in meditation a hundred times, has watched the small flare of restlessness that precedes it and the small deflation that follows. None of that watching, by itself, has stopped the reach. This is not a failure of insight. It is a fact about how repeated action shapes the instrument that produces future action, and both a two-thousand-year-old psychological system and a current neuroscience literature have independent, oddly convergent accounts of why.

The classical account uses the term samskara: a subliminal impression, a residue or imprint left in the mind by an action, perception, or emotional event. A samskara is not the memory of the event. You can forget entirely that you once did something and still carry the imprint it left, the readiness it created for the mind to move the same way again.

The modern account uses different nouns, cortico-striatal weight, stimulus-response association, procedural memory trace. But is describing the same basic phenomenon: repetition leaves the machinery changed in a way that outlasts the deliberate intention that started it. The changed machinery then produces behaviour with less and less need for deliberate intention at all.

This piece sets the two accounts side by side. Classical Indian psychology (drawing mainly on Yoga philosophy, Sankhya, and Vedanta, with a companion but distinct treatment in Buddhist Abhidhamma) built a detailed model of samskara, vasana. Karma centuries before anyone had a microscope fine enough to look at a synapse. Modern habit neuroscience built its own model from lesion studies, dopamine recordings, and behavioural pharmacology, with no reference to the Indian material at all.

The two converge on a strikingly similar functional picture and diverge on what that picture ultimately means. Getting the comparison right requires being honest, at every step, about which claims are experimentally testable and which are first-person, doctrinal, or metaphysical.

The classical tradition is not making a set of failed scientific predictions that we can now grade. It is a different kind of account, built from sustained introspective observation and organised around soteriological aims, and it deserves to be read on those terms rather than flattened into either superstition or crypto-neuroscience.

The classical model

Samskara as trace

In the Yoga Sutra and its classical commentaries, the mind (citta) is treated as a substance capable of taking on modifications. Every act of cognition, perception, or emotional response leaves behind a subtle residue once the active mental event has passed. That residue is the samskara.

Vyasa's commentary on the Yoga Sutra describes samskaras as the seeds from which future mental activity of a similar kind arises, and the relationship is explicitly circular: an experience produces a samskara. A samskara, when conditions are right, produces an experience or action of the same general shape, which in turn deposits a further samskara. The loop reinforces itself.

Sankhya psychology, which shares much of its vocabulary with classical Yoga, treats samskaras as belonging to the subtle body, the substrate that carries psychological dispositions independently of any single perception or thought.

Two features of the classical description are worth holding onto because they will matter later when comparing with neuroscience. First, samskaras are described as latent rather than always active. They lie dormant in the mind and are activated by suitable conditions, the way a seed lies in soil until moisture and warmth trigger germination. Second, samskaras are cumulative and additive rather than singular. No individual samskara fully determines behaviour on its own; it is the accumulation of many similar samskaras, laid down over repeated instances of a similar action or reaction, that produces a noticeable behavioural tendency.

Vasana as tendency field

That aggregate is named separately: vasana. The word derives from a root meaning to dwell or to perfume. The standard image invoked in commentarial literature is of a cloth that has held incense and continues to carry its scent long after the incense itself is gone. A vasana is the residual scent, the persistent tendency-field built from the accumulation of many individual samskaras of a similar type.

Where a samskara is a discrete trace left by one action or experience, a vasana is closer to what a psychologist would call a trait-level disposition. A standing tendency toward anger, toward a particular craving, toward a characteristic way of interpreting social slights, built from the statistical residue of many prior instances rather than any single one.

The Yoga Vasistha, a later Vedantic text organised almost entirely around this concept, treats vasana as the operative mechanism binding a person to a repeated pattern of experience. Treats the deliberate thinning of vasana as close to the entire practical content of the text's teaching. The relationship between samskara and vasana in these texts is roughly part to whole: samskaras are the discrete deposits, vasana is the accumulated field of tendency they compose. It is vasana, not any individual samskara, that shows up as a stable trait a person or the people around them can recognise and name.

Karma as causal action

Karma is frequently mistranslated in casual usage as fate or destiny. Its more precise sense in this literature is volitional action together with the causal residue that action leaves behind. Karma is the doing; samskara is the trace that doing deposits; vasana is the accumulated tendency built from many such traces. The doctrine of karma-phala, the fruiting of action, is the traditional claim that this causal residue eventually ripens into experienced consequence, sometimes within the same life and, in the wider classical cosmology, sometimes across lives.

It is worth separating two layers of claim here, because a modern reader can easily accept or reject them as a single package when they are not epistemically equivalent.

The first layer, that repeated volitional action shapes the disposition of the agent who performs it and biases the probability and ease of future similar action within a single lifetime, is a psychological claim that has a reasonably direct behavioural analogue and is, in a loose sense, testable. It predicts exactly the kind of stimulus-response strengthening that habit researchers now study directly.

The second layer, that unresolved karmic residue carries forward across lives and determines circumstances in a future rebirth, is a metaphysical claim embedded in a cosmology of rebirth that sits outside what any current experimental method can confirm or falsify. This essay is concerned with the first layer, because that is where the classical and modern models can be compared on shared ground.

The second layer is described here as the tradition itself describes it, as a first-person doctrinal claim resting on textual authority and reported meditative perception, not as an established empirical finding, and not as something to be dismissed either. It sits outside the scope of what this comparison can adjudicate.

The riverbed metaphor

Commentarial literature across several of these traditions reaches for a shared image to describe how repetition deepens a samskara into a stable vasana. Water flowing over soft ground does not need a channel to begin with. But the first passage of water leaves a shallow depression, and each subsequent flow follows that depression rather than cutting a fresh path, deepening it further with every pass. Given enough repetition the depression becomes a fixed channel, and water will follow it even when a flatter, easier path lies immediately alongside.

The metaphor is doing real conceptual work rather than merely decorating the point. It captures that the direction of least resistance for future action is not fixed in advance but is carved by the very actions that then travel it, that the channel deepens by positive feedback, that a shallow channel is comparatively easy to redirect while a deep one resists redirection even when the redirection is only a slight lateral push. That the channel persists as a structural feature of the terrain even during a season when no water happens to be flowing through it, which is the classical answer to why a samskara can lie dormant for years and still reassert itself the moment conditions recur.

The modern habit loop

Cue routine reward

Contemporary habit research, following work by psychologists such as Wendy Wood and neuroscientists such as Ann Graybiel, describes habitual behaviour as organised around a three-part loop. A cue in the environment or internal state, a routine behaviour triggered by that cue. A reward that follows the routine and reinforces the association between cue and routine. Early in learning, the behaviour is goal-directed: the organism performs the action because it currently values the outcome and currently believes the action produces that outcome.

With enough repetition under stable reward conditions, control shifts to a more automatic mode, where the cue itself triggers the routine with little or no reference to current goal value at all.

The classic laboratory demonstration of this shift is outcome devaluation. An animal trained extensively on a lever-press for food will, after the food is paired with illness or after being fed to satiety, continue pressing the lever even though the outcome has just been made aversive or irrelevant, which is behavioural proof that the action is no longer being selected by a live evaluation of its consequence.

From goal directed to habitual control

The circuit-level story for this shift centres on the striatum, a subcortical structure with distinct subregions implicated in distinct modes of control. Lesion and inactivation studies in rodents, associated especially with the work of Henry Yin, Barbara Knowlton. Bernard Balleine, show that the dorsomedial striatum is required for goal-directed action selection, the mode sensitive to outcome devaluation, while the dorsolateral striatum becomes progressively more necessary as training proceeds and behaviour becomes habitual, the mode that persists despite devaluation.

The standard summary is that control shifts from a dorsomedial, action-outcome system to a dorsolateral, stimulus-response system as a behaviour is repeated under stable conditions. This dorsomedial-to-dorsolateral account should be treated as the current best working model built from converging animal-lesion and human-imaging evidence rather than as a fully settled account of every case of habitual behaviour in humans. The underlying circuitry is more distributed and the human evidence, largely from functional imaging correlations rather than causal manipulation, is necessarily less direct than the rodent lesion work.

With that caveat in place, it remains the most carefully validated bridge currently available between a psychological description of habit and an anatomical one.

Dopamine and reward prediction error

A second empirical thread concerns dopamine. Wolfram Schultz's recordings in primates showed that midbrain dopamine neurons fire not simply in proportion to reward received but in proportion to reward prediction error, the difference between an outcome and what was expected. Early in learning, an unexpected reward produces a strong dopaminergic burst. As learning proceeds and the reward becomes predictable, the burst shifts earlier in time, onto the cue that reliably predicts the reward, and firing to the now-expected reward itself diminishes.

This reward prediction error signal is widely treated within computational neuroscience as an implementation of a temporal-difference learning rule, the same class of algorithm used in reinforcement learning. It offers a mechanistic account of exactly the transfer classical descriptions gesture at without formal detail. Value and salience migrate from outcome to the cue that precedes it, which is part of why a well-established cue becomes so effective at triggering the routine on its own, independent of any deliberate evaluation of the outcome at the moment of triggering.

The samskara model and the habit-circuit model are describing convergent phenomena from opposite directions, one built from centuries of disciplined introspective report organised for the purposes of practice, the other built from controlled lesion, pharmacology. Electrophysiology organised for the purposes of prediction and mechanism. Where they line up is instructive; where they do not, equally so.

Predictive processing and stubborn priors

Predictive processing frameworks, most prominently the active inference formulation associated with Karl Friston, model the brain as a hierarchy of generative models that continuously predict incoming sensory and interoceptive data and update those predictions when prediction error arrives, weighted by how confident, or precise, the system currently is in each level's model. Habitual behaviour fits naturally into this picture as a low-level policy, a tightly specified action sequence, assigned very high precision.

High precision at a given level means prediction error at that level is treated as reliable and is allowed to drive fast, low-latency motor output, while simultaneously being treated as unlikely to need revision, because the system has, in effect, staked a great deal of confidence on that level's model being correct.

This is a coherent functional explanation for why habits are so hard to interrupt through deliberate top-down effort alone. The higher-level goal representation, however clearly it disagrees with the habitual policy, is competing against a lower-level prior that has been assigned high confidence precisely because it has proved reliable across a large number of past instances. High-confidence priors are, by construction, resistant to being overridden by a single act of conscious disagreement.

This framing should be held with appropriate caution. Predictive processing is itself an active, contested research programme rather than an established finding, useful as an organising vocabulary and generator of specific predictions but not yet validated to the point of being treated as settled neuroscience.

Used carefully, though, it offers a genuinely useful bridge concept for describing what the riverbed metaphor was reaching for without a formal notion of precision. A shallow, recently formed samskara corresponds to a low-precision prior, easily revised by contrary information or a single strong counter-instance, while a deep, long-reinforced vasana corresponds to a high-precision prior, one that keeps generating the old behavioural prediction even after the practitioner has, at the level of explicit belief, thoroughly revised their view of the action's value.

The felt experience many long-term meditators report, of seeing an urge clearly, disagreeing with it completely. Still watching the body begin to execute the old routine, maps with unusual precision onto a system in which a high-level belief has changed while a low-level, high-precision policy has not yet been given enough disconfirming, precision-weighted evidence to update.

Where contemplative practice intervenes

Pratipaksha bhavana

The Yoga Sutra offers a specific technique for working directly on an established negative tendency: pratipaksha bhavana, the deliberate cultivation of an opposing tendency to counteract a harmful one. Patanjali's text states that when troubled by unwholesome thoughts, one should cultivate their opposite. The commentarial tradition elaborates this as an active, repeated practice rather than a single act of will. The practitioner does not merely suppress the unwanted impulse but deliberately rehearses an incompatible response until that response itself becomes a competing, and eventually a preferred, channel.

This maps closely onto what modern behaviour-change research calls competing response training or habit substitution. Rather than attempting to erase an established stimulus-response link through sheer inhibition, the intervention builds a second, incompatible stimulus-response link to the same cue and lets the two compete, with the newer link needing repeated reinforcement of its own before it can reliably win.

Vipassana noting

A second and functionally different lever is offered by vipassana practice. Vipassana is the Pali term for insight, referring to a family of practices, most systematically developed within the Theravada Abhidhamma tradition, aimed at direct, moment-by-moment observation of experience as it arises, rather than at cultivating a competing behaviour. In its noting form, the practitioner learns to register the arising of an urge, a sensation, or an emotional flare with a brief, non-elaborative label at the moment it appears, before it has automatically cascaded into the habitual routine.

The claimed mechanism of action here is different from pratipaksha bhavana. Rather than installing a competing response, noting practice inserts a moment of explicit registration into a sequence that would otherwise proceed automatically, at the specific point where craving is arising but the motor routine has not yet been triggered.

Judson Brewer's clinical research on craving and mindfulness training, conducted with smokers and other populations with strong habitual cravings, is relevant evidence here. Participants trained to observe craving with sustained, non-reactive attention showed reduced craving intensity and, in several trials, better cessation outcomes than standard behavioural counselling, which is a genuine empirical data point in favour of the mechanism, though it should be read as promising clinical evidence from a still-developing literature rather than as a full mechanistic confirmation of what noting practice is doing at the level of striatal or prefrontal circuitry.

Nirodha and non reinforcement

The Abhidhamma frames the deepest lever as nirodha, cessation, specifically the cessation of a chain of conditioned arising once its supporting conditions are no longer supplied. The doctrine is explicit that nirodha is not primarily achieved by forceful suppression of an arisen state but by withdrawing the conditions, principally craving and clinging, that would otherwise carry one link of the chain into the next.

Practically, this converges with the modern behavioural principle of extinction: a conditioned response weakens when it is repeatedly triggered without the reinforcement that originally established it. An animal that presses a lever for food will, over repeated unrewarded presses, press less. A person who registers the urge to check the phone without following through into the checking, repeatedly, will over time see the intensity and frequency of that urge diminish.

The classical tradition and the modern behavioural literature agree on something with genuine practical weight: neither treats this weakening as the erasure of the underlying trace. The Abhidhamma explicitly holds that a samskara, once laid down, is attenuated and rendered dormant by right practice rather than deleted outright, which is precisely mirrored in the modern finding that extinguished responses are suppressed rather than unlearned, a point developed further below.

Dependent origination as a finer grained model

The Abhidhamma's account of habit formation operates at a much finer temporal grain than either the samskara-vasana model or the modern cue-routine-reward loop, through the doctrine of paticca samuppada, dependent origination, a chain describing how each moment of conditioned experience gives rise to the next.

In its classic twelve-link formulation the chain runs from ignorance through mental formations, consciousness, name-and-form, the six sense bases, contact, feeling, craving, clinging, becoming, birth. Finally ageing and death, with the whole sequence sometimes read across a single lifetime and sometimes, in the traditional cosmological reading, across successive lives.

For the purposes of habit work, the operative and most frequently cited segment is shorter. Contact with a sense object gives rise to feeling, feeling gives rise to craving if it is pleasant or aversion if it is unpleasant, craving gives rise to clinging. Clinging gives rise to the action, becoming, that both expresses and reinforces the tendency.

The pivot the tradition repeatedly identifies as the most tractable point of intervention is the link between feeling and craving: contact and the raw feeling-tone it produces are described as largely unavoidable and pre-volitional. But the movement from feeling into craving is where the chain first becomes susceptible to deliberate attention. Interrupting it there, before craving hardens into clinging, is described as requiring far less effort than trying to stop the chain once clinging has already produced the compulsion to act.

It is worth being exact about what kind of claim this is. Paticca samuppada is presented within the tradition as a description arrived at through sustained first-person observation of one's own arising and passing mental events, organised for a soteriological purpose, the reduction of suffering through the interruption of a specific link, rather than as a report of an externally measured neural sequence.

There is a loose structural resonance with predictive processing's own moment-by-moment account of how sensory contact generates an affectively toned prediction that then drives further inference and action. Some contemplative-neuroscience researchers have proposed exploratory mappings along these lines. But that resonance is a hypothesis for future interdisciplinary work, not a validated finding. The twelve-link chain as a whole, particularly its links concerning birth and future becoming across lives, remains a doctrinal and phenomenological model rather than an empirically confirmed neural mechanism.

Presenting it as neuroscience that has simply not yet been measured would overstate what is currently known. Presenting it as mere metaphor would understate how carefully and specifically it was derived from disciplined introspective method within its own tradition.

Relapse spontaneous recovery and what persists

Both traditions converge on a claim with substantial practical consequence: a diminished habit is not the same as a deleted one. The mechanism of diminishment leaves the original trace intact in some attenuated or dormant form. The Abhidhamma and the broader Yoga-Vedanta literature describe attenuated samskaras as capable of reactivating under sufficiently strong triggering conditions, most classically under stress, fatigue, or a return to the original environmental cue, even after long periods of apparent absence.

The modern extinction-learning literature, developed largely through the work of Mark Bouton et al., documents the identical pattern with considerable experimental precision. Extinction training, repeatedly presenting a conditioned cue without its reinforcer, reliably reduces the conditioned response. But the original association is not unlearned; a new, competing inhibitory association is learned that suppresses expression of the old one. This is demonstrated by three well-replicated phenomena. Spontaneous recovery: an extinguished response can reappear after the mere passage of time, with no new exposure to the cue at all.

Renewal: an extinguished response reappears if the organism is moved out of the specific context in which extinction training occurred, back into the context in which the original habit was formed, or into any novel context. Reinstatement: an extinguished response can be triggered back into full strength by a single unsignalled exposure to the original reinforcer, even years after apparently successful extinction.

The practical implication for anyone doing habit work, contemplative or behavioural, is the same regardless of which vocabulary one prefers. A period of successfully not performing a habitual action, even a long one, is evidence of active suppression by a competing process, not evidence that the underlying tendency has been erased. Relapse under stress, after a change in environment, or after a long gap is not a sign that the practice failed; it is close to the default prediction of both models. Treating it as a moral failure rather than an expected feature of how conditioned tendencies actually behave is itself a significant obstacle to sustained change.

A practice protocol for working with a habit

The following protocol combines classical technique and modern behavioural method on a single target habit. Is written for someone willing to work on one specific, narrowly defined behaviour for a period measured in months rather than days.

1. Define the target with precision. Not "I want to be less distracted" but "I check my phone within the first sixty seconds after a meeting ends." A vague target cannot be noted. A vague target cannot be measured against a baseline.

2. Run a baseline noting period of one to two weeks before attempting any change. Each time the behaviour occurs, or the urge for it arises, note the cue that preceded it, the felt quality of the urge (its intensity, its location in the body, its duration), and whatever followed the action. This period is deliberately not yet an intervention; its purpose is to make the cue-craving-routine-reward structure visible, since neither classical practice nor modern behaviour change works well against a pattern that has not first been mapped.

3. Redesign the environmental cue where possible. If the phone sits on the table in view, move it out of the room or into a drawer during the relevant window. This is not a contemplative technique; it is a straightforward application of the finding that habitual behaviour is strongly cue-triggered and that removing or altering the cue reduces triggering frequency with very little effort, buying time for the other two interventions to do slower work.

4. Practise noting at the moment of urge. The specific skill is catching the urge after it arises but before the routine executes. Applying a brief, non-elaborative label (wanting, restlessness, urge) without immediately following either the urge or a forced suppression of it. The aim in early weeks is simply to lengthen the gap between urge and action, even by a few seconds; that gap is where any subsequent choice becomes possible at all.

5. Install pratipaksha bhavana, a specific, pre-planned opposite action to perform in the gap created by noting. It should be concrete and rehearsed in advance, not invented in the moment: three slow breaths with the hands folded, or reaching for a specific object instead of the phone. The opposite action needs its own repetition to become competitive with the old routine. Expect it to feel effortful and unnatural for a period of weeks, since it is a new, shallow channel competing against an old, deep one.

6. Practise in spaced, repeated sessions rather than relying on a single burst of willpower. Widely cited survey data from Phillippa Lally et al. found that the time for a new behaviour to reach a stable, automatic asymptote varied enormously across individuals and behaviours, with a median around sixty-six days and an observed range from about eighteen to over two hundred and fifty days. That range should be read as an approximate empirical anchor rather than a firm prediction for any specific habit. Simpler behaviours in stable contexts tend toward the shorter end, complex or emotionally loaded behaviours toward the longer end.

7. Review weekly rather than daily. Daily self-assessment on a habit this granular tends to produce either premature discouragement or false confidence; a weekly review of the noting log gives enough data to see trend rather than noise.

8. Expect and plan for relapse rather than being surprised by it. Identify in advance the two or three conditions most likely to trigger reinstatement, commonly acute stress, fatigue, travel, or return to an old physical environment. Treat a lapse under one of those conditions as confirmation of the model rather than as evidence of failure. Resume noting and the opposite action immediately rather than waiting to feel motivated again.

None of these six methods is guaranteed and none is a substitute for the others. The classical tradition and the modern literature agree, from different premises, that durable change comes from combining a reduction in triggering (cue redesign, non-reinforcement) with the active construction of a new, competing channel (pratipaksha bhavana, habit substitution), sustained over a timeframe measured in months, with relapse treated as data to act on rather than as proof that the underlying model has failed.

For practitioners

Pick one habit, not three. The temptation to run this protocol against every unwanted pattern at once dilutes the attention any single pattern needs to become visible. The baseline noting week works better as a narrow, well-lit observation than as a broad survey.

Do the baseline week honestly, including the instances that embarrass you. A noting log with gaps in it, from moments you did not want to look closely, is a log that will underestimate frequency and mislead your later review.

Treat the environmental redesign step as the cheapest win available and take it immediately; there is no reason to wait until your noting practice matures before moving the phone out of the room. Cue removal and internal practice are not competing strategies, they are complementary. The classical tradition's emphasis on internal cultivation should not be read as a reason to neglect the much simpler external lever sitting next to it.

When you feel the old urge return after weeks of apparent quiet, resist the interpretation that the work has failed. Both the classical account of dormant samskara and the modern account of spontaneous recovery predict exactly this. What distinguishes someone who eventually changes the pattern from someone who does not is rarely the absence of relapse. It is the speed and steadiness with which noting and the opposite action are resumed after a lapse, without the added layer of self-recrimination that tends to extend the lapse into a longer collapse.

Hold the metaphysical claims of the tradition, particularly those concerning karma across lives and the ultimate cessation of the entire chain of dependent origination, separately from the practical claims you are testing on yourself. You do not need to adjudicate rebirth to test whether noting practice lengthens the gap between urge and action for you specifically. Conflating the two only invites you to reject a workable practical technique because you are unconvinced by a metaphysical claim it does not actually require, or to accept a metaphysical claim uncritically because a related practical technique happened to work.

Finally, keep your expectations of timeframe realistic and specific to the behaviour rather than borrowed wholesale from a headline figure. A well-defined, low-stakes behaviour in a stable context may shift meaningfully within a few weeks of consistent practice. An emotionally loaded, high-stakes pattern tied to stress or identity may take the better part of a year of sustained, spaced work, with visible relapse along the way. Both outcomes are consistent with everything the classical model and the modern one predict about how a deep channel is reshaped.