1. From instrument to habitus

Theological reflection on artificial intelligence (AI) has, until recently, been dominated by the register of applied ethics. In monographs, policy statements, and ecclesial position papers, leading questions have concerned algorithmic bias, the moral status of autonomous systems, the exploitation of data labour, and the erosion of privacy within what Shoshana Zuboff describes as surveillance capitalism and the attention economy.Footnote ¹ These inquiries are necessary. They disclose the material, social, and political costs of a computational infrastructure that increasingly mediates human interaction and institutional judgement. Yet they often share a common limitation: they frame the algorithm primarily as an instrument that stands over against the moral agent, awaiting regulation, resistance, or correct use.

This instrumental approach risks missing a second mode of algorithmic power. Algorithms are not only used. They are also inhabited. Through continuous exposure and feedback, platforms become environments that train perception and expectation, shape attentional habits, and guide affective orientation. In Thomistic terms, such systems do not merely place objects before the will. They influence the preconditions of willing by forming dispositions and inclinations that make certain acts easier, likelier, and more ‘natural’ over time. In that sense, they function as a species of habitus,Footnote ² not as virtues or vices in themselves but as an ecology of repeated prompts, cues, and reinforcements that gradually bends the field of practical reason.

If algorithmic mediation operates at this depth, then a purely ethical framing is insufficient. The question becomes not only what one should do with algorithms but also what algorithms are doing to the agent. This article therefore pivots from external moral evaluation to a dogmatic locus that names formation at the level of desire: sanctification. In classical theology, sanctification is not reducible to moral achievement. It is the Spirit’s gradual reordering of the person towards the summum bonum through the infusion of charity and the healing of disordered loves. If, as Aquinas argues, the invisibilis missio of the Holy SpiritFootnote ³ is known by its effects in the interior life, above all the reordering of affections, then any mass-scale system that trains attention and desire at the level of imagination and appetite becomes a significant theological interlocutor.

My central thesis is that contemporary predictive and recommender systems can be read as a counter-formation that structurally imitates aspects of moral and spiritual formation while inverting its end. I describe this as a ‘counter-pneumatology’ in a strictly analogical and structural sense, not as an ontological claim about algorithmic agency. The point is not that algorithms are agents that rival the Spirit but that they can operate as a rival ecology of habituation. By anticipating engagement and shaping the user’s field of salience, such systems can produce a closed loop of repetition that reinforces prior patterns of desire and narrows the horizon of practical judgement. Where the Spirit’s mission opens the will towards charity and communion, algorithmic habituation can enclose the will within an economised circuit of attention oriented towards extraction.

The term ‘counter-pneumatology’ is therefore employed analogically and with deliberate restraint. The comparison concerns structural resemblance in the economy of formation, not ontological parity between divine and technological agency. In Thomistic terms, the analogy is one of ordered attribution rather than univocity. To prevent category confusion, I maintain a consistent distinction between three levels of claim throughout the article. The first is ontological: the Holy Spirit is a divine Person whose indwelling is the source of uncreated and sanctifying grace, and this grace cannot be placed in competition with man-made mechanisms. Throughout what follows, I preserve the Thomistic distinction between the Person of the Holy Spirit and the created effect of sanctifying grace (gratia creata), while recognising that the former is the principle of the latter. The second is teleological: digital systems are ordered towards objective functions shaped by institutional incentives, whereas sanctification is ordered towards communion with God through infused charity. The third is phenomenological and pedagogical: both sanctification and platform habituation exhibit temporally extended patterns of repetition that shape inclination prior to explicit deliberation. When I speak of ‘counter-pneumatology’, the claim operates strictly on this third level. It designates a structural resemblance in the economy of formation, not an attribution of spiritual agency to technical artefacts, and it should be read as a diagnostic term for rival habituation rather than as a metaphysical proposal.

The Spirit’s invisible mission names the divine mode of interior elevation and reordering of love; algorithmic systems, by contrast, are artefacts embedded within created causality. The analogy lies in the temporal structure of habituation: both operate through patterned repetition that shapes inclination prior to explicit deliberation. The disanalogy is absolute at the level of being and intention. The Spirit moves, as the principle of uncreated grace and the source of its created effects in the soul, towards communion with the summum bonum; the algorithm operates as a human-made artefact ordered towards optimisation functions defined by institutional incentives. Clarifying this asymmetry is essential. The claim is not that algorithms rival God in agency but that certain configurations of digital mediation can structurally imitate the dynamics of formation while redirecting them towards ends other than charity.

The argument unfolds in four movements. First, I retrieve Augustine’s notion of pondus and Aquinas’s distinction between actus humanus and actus hominis to describe how platform environments train pre-reflective desire. Second, I examine the exteriorisation of memory and argue that algorithmic profiling produces a ‘Counter-Imago’, a functional simulacrum that tends towards closure by projecting the past forward as fate. Third, I propose that sanctification in an algorithmic environment can be described as a restored capacity for ‘holy hesitation’, a Spirit-formed dilatio that reopens the interval between impulse and assent. Finally, I locate a decisive counter-formation in Eucharistic temporality, where the Church inhabits a time shaped by promise rather than prediction. In this way, the doctrine of sanctification offers a theological grammar for naming and interrupting an ordo extractionis that otherwise presents itself as neutral convenience.

2. The metaphysical and existential turn in AI theology

To argue that algorithms function as a ‘counter-pneumatology’ is to intervene in a field that has already matured significantly beyond the boundaries of applied ethics. As recent scholarship demonstrates, theological engagement with AI now encompasses robust metaphysical, meta-ethical, and existential inquiries. It is necessary, therefore, to situate the present argument within this broader dogmatic landscape in order to distinguish the specific contribution of a Thomistic theology of sanctification. Significant work has addressed the ontological status of artificial agents. Theologians such as Noreen HerzfeldFootnote ⁴ have examined the analogy between AI and the imago Dei, arguing that AI serves less as a replication of human interiority than as a mirror reflecting the relational and anthropological assumptions embedded within its design. In parallel, within existential and eschatological theology, scholars such as Michael Burdett and Brent WatersFootnote ⁵ have critically assessed the transhumanist currents associated with technological development, showing how narratives of technological progress can function as secularised soteriologies that imitate, yet distort, Christian accounts of hope and fulfilment.

Furthermore, within meta-ethics, the renewed appeal to virtue ethics in the philosophy of technology, articulated by thinkers such as Shannon Vallor and extended within Christian ethics by Brian Brock,Footnote ⁶ has foregrounded the formative impact of technological mediation on moral character. Rather than treating technology as a neutral tool awaiting moral application, this body of work demonstrates that technological environments participate in the cultivation, deformation, and redirection of habitus. In doing so, it establishes that the moral significance of technology lies not only in its uses but in its capacity to shape the dispositions through which agents perceive, judge, and act.

However, a distinct dogmatic gap remains. While the metaphysical literature focuses heavily on the nature of the machine (Can it think? Does it have a soul?), and the existential literature focuses on the teleology of history (Will it save us?), there is a need for a specific account of the mechanism of formation in the present. The question is not only whether AI is a ‘person’ but how it acts upon the human person as a rival habitus.

The argument for a ‘counter-pneumatology’ does not displace these metaphysical inquiries but complements them by shifting the locus to pneumatology and sanctification. If Herzfeld focuses on the Image (the Son), this article focuses on the Breath (the Spirit). The danger of predictive AI is not merely that it creates a false idol of the human (the ontological problem) but that it instantiates an extractive economy of attention and desire, the pneumatic problem. By grounding this critique in Aquinas’s doctrine of the invisibilis missio, we move beyond the question of whether the machine is ‘alive’ to the urgent pastoral question of why responsiveness to grace can be attenuated under extractive regimes of attention. The remainder of this article develops that pneumatological turn in four movements. I begin, in the next section, by examining the phenomenology of predictive systems, for only once the formative mechanism is described with precision can the theological grammar of sanctification be brought into critical relation with it.

3. The phenomenology of prediction: salience, compression, and the pressure on deliberation

To describe algorithmic formation with sufficient precision, it is important to clarify what is and is not being claimed. The argument concerns contemporary recommender and ranking systems that optimise engagement or related proxy metrics by learning from user behaviour over time. Such systems include collaborative filtering, content-based ranking, and reinforcement-like feedback loops in which repeated exposure and interaction shape subsequent recommendations. The argument does not depend on claims about machine consciousness, agency, or personhood. It concerns the structural effect of optimised curation on attention, imagination, and habitual desire.

It is important to delimit the scope of this analysis. I am not proposing a universal account of all algorithmic systems, nor suggesting that every digital mediation functions identically. Recommender architectures differ in objective functions, training data, and institutional context, and users retain varying degrees of reflexive awareness and resistance. The argument concerns a specific class of predictive systems that optimise for engagement, retention, or monetisable interaction through iterative feedback loops. Nor is the claim that such systems determine the will or eliminate responsibility. Rather, the narrower thesis is that under certain design conditions they can, in practice, condition salience and compress deliberative intervals in ways that exert formative pressure on desire. The theological analysis that follows is therefore analogical and diagnostic, not totalising.

Within such environments, algorithmic mediation often pre-structures choice by pre-selecting salience. Before the intellect explicitly judges a candidate object of attention, the system has already shaped the field in which objects appear as worthy of attention at all. Through signals such as dwell time, scroll behaviour, interaction traces, and the dynamics of one’s social graph, platforms estimate the probability of engagement and adjust what is presented next.Footnote ⁷ This operation occurs at a temporal scale far shorter than reflective deliberation. A new prompt, image, or video becomes available as soon as attention begins to loosen, with the result that the transition from impulse to stimulus can become frictionless.

It would be mistaken, however, to construe this dynamic as unilateral imposition. Algorithmic habituation is co-constituted through repeated acts of attention, curiosity, and return. Every refinement of the predictive model presupposes voluntary interaction, however minimal. The user is not merely acted upon but participates in the iterative stabilisation of preference through micro-assents that accumulate over time. In Thomistic terms, even habituation formed under environmental pressure still engages the will, though often at the level of actus hominis rather than fully reflective actus humanus. The ecology of optimisation and the appetite of the subject intersect in a feedback circuit in which design incentives and human desire reinforce one another. This mutuality does not eliminate structural asymmetry in power or knowledge, but it prevents a deterministic account of formation. Algorithmic systems condition salience: they do not extinguish agency. The theological concern, therefore, is not technological fatalism, but the gradual normalisation of reactive consent.

This compression can be rendered with greater conceptual precision by placing it in dialogue with recent accounts of temporality and evaluation in late modernity. Social acceleration, as described by Hartmut Rosa,Footnote ⁸ names not only faster communication but the systemic shortening of the temporal horizons within which meaning can stabilise, thereby increasing the frequency of reactive adaptation and reducing the space for resonant, purposive engagement. Charles Taylor’s account of ‘strong evaluation’Footnote ⁹ likewise clarifies that mature agency does not consist in selecting among options at speed, but in interpreting and ranking goods within a horizon of significance. Algorithmic curation does not negate strong evaluation, but it can render it more difficult by thickening the stream of stimuli and narrowing the time available for reflection on ends. Under these conditions, dilatio is not merely ascetical slowness; it is the recovery of the temporal conditions under which strong evaluation becomes practically possible.

In Thomistic psychology, this matters because practical reason depends upon a minimally stable interval in which the intellect can apprehend, judge, and propose an end to the will. Aquinas distinguishes between the sensitive appetite, moved by passions through sense and imagination, and the rational appetite, the will, moved towards a good as apprehended by the intellect.Footnote ¹⁰ Algorithmic environments do not remove the intellect, nor do they literally bypass it. Rather, they exert pressure upon it by saturating imagination and by compressing the temporal space in which judgement matures. The system operates by keeping the sensitive appetite continuously supplied with phantasms, that is, imagination-laden sensory representations of high salience, so that cognition is repeatedly scaffolded by what has already been selected as engaging. This produces a practical effect that can be described using Augustine’s language of dilatio. There is a narrowing of the interval in which the soul can distinguish between transient impulse and a good worthy of sustained pursuit.Footnote ¹¹ The tempo of presentation shortens the arc of desire, reducing the time in which the intellect can evaluate ends and the will can consent with freedom rather than momentum. In that sense, algorithmic habituation can corrode the conditions of mature deliberation, not by cancelling rationality, but by disciplining the agent into a more reactive mode of attention.

Augustine’s dictum, ‘my weight is my love; by it I am borne wherever I am borne’,Footnote ¹² is diagnostic here. The algorithm does not merely record pondus as data. It can participate in redistributing pondus by amplifying what captures attention and by repeatedly returning the subject to the same family of affects. This is not deterministic, but it is formative: the more the imagination is trained on the sensational and immediate, the more the appetite becomes at home there. A habit of restlessness can become easier than a habit of sustained attention.

At this point, the analogy to prevenience becomes intelligible. In classical theology, grace is prevenient because it precedes the free act of the will, healing nature and preparing it for the good.Footnote ¹³ Algorithmic systems can imitate the temporal structure of ‘going before’ by anticipating likely engagement and placing stimuli in advance of explicit choice. The difference is teleological: where prevenient grace is gift ordered towards charity and communion, predictive curation is oriented towards the reproduction of engagement and the extraction of value. It therefore tends towards a regress of desire, in which the future is shaped as a statistically refined repetition of the past, unless interrupted by counter-formative practices.

This ordo extractionis is not only psychological. It is materially and socially instantiated in infrastructures that distribute costs and benefits asymmetrically. The training and deployment of large-scale AI systems depend on extractive supply chains, energy-intensive computation, and labour arrangements that are frequently rendered invisible at the point of consumption.Footnote ¹⁴ Moreover, ranking and search systems can encode and amplify structural injustice by translating historical inequalities into apparently neutral outputs, thereby shaping not only what is seen but who is rendered credible, suspect, or disposable.Footnote ¹⁵ These realities matter for sanctification because the formation of desire is never purely interior. Attention is trained within economies of power, and love is habituated within regimes of recognition.

The Spirit’s reordering of desire, therefore, has an irreducibly social dimension: charity reforms not only personal appetites but also the patterns of attention and valuation by which neighbours are either apprehended as persons or reduced to functions. The ordo extractionis is not merely an economic arrangement; it is also a deformation of the I–Thou relation. Under conditions of pervasive datafication, the neighbour is encountered first as a data point, a statistical profile, or an engagement surface, and only derivatively, if at all, as the singular one whose face places me under obligation. Where charity encounters the neighbour as irreducibly other, predictive mediation tends to pre-interpret the other as a probable response. Sanctification as the reordering of love must therefore include the restoration of the neighbour from datum to person, a restoration enacted sacramentally in the Eucharistic body and politically in the concrete practices of justice that refuse to reduce persons to the measurable.

If so, the question of sanctification is not an inward piety detached from politics. It is the Spirit’s reordering of love in a world where attention is continuously commodified. A theological account of formation must therefore take seriously both the interior dynamics of desire and the exterior structures that profit from its capture.

4. Memoria and the Counter-Imago: profiling, closure, and the question of repentance

The theological stakes become sharper when algorithmic formation is considered through Augustinian anthropology. In De Trinitate, Augustine locates the image of God in the triad of memoria, intellectus, and voluntas.Footnote ¹⁶ Memory here is not merely storage. It is an interior depth that sustains personal continuity across time and renders the self present to itself. Memory is also the condition for conversion, because repentance requires that the past be remembered truthfully while the future remains open to transformation.

Augustine’s triad in De Trinitate clarifies why this matters. Memoria, intellectus, and voluntas are not three detachable faculties but mutually interior moments of one living act of self-presence. Memoria is not a warehouse of impressions; it is the abiding depth in which the self holds itself across time, within which the intellect judges and the will loves. It is therefore intrinsically ordered towards truth and beatitude, because the memory that grounds identity also carries the implicit orientation towards God as the ultimate good. Data retention, by contrast, abstracts traces from this interior unity. It stores behavioural residues without preserving the inward synthesis by which the person understands and appropriates them. What is lost in the transition from memoria to data is precisely the capacity of the self to stand within its past as a judging and loving subject. Repentance presupposes this interior depth: it is the act by which memoria, illumined by intellectus and moved by voluntas under grace, re-narrates the past in light of a promised future. Where data projects the past forward as statistical continuity, memoria under grace opens the past towards conversion. In that difference, the theological meaning of futurity becomes visible.

Algorithmic environments affect this structure by exteriorising and operationalising memory. A technical clarification is needed at this point. Many recommender systems incorporate mechanisms that introduce novelty, serendipity, or exploration, and their objective functions can include diversity constraints rather than pure reinforcement of prior preferences. Yet these features do not remove the basic structural tendency that concerns me here. Even when novelty is introduced, it is typically selected as novelty for this user, and the exploration is instrumentally governed by downstream measures of retention, satisfaction, or monetisable interaction. The question is, therefore, not whether all recommendation produces closure without remainder but whether the overall architecture tends to privilege iterated reinforcement over open-ended transformation. The Counter-Imago names that tendency towards probabilistic continuity when the past is repeatedly operationalised as the primary guide for what the self is likely to want next. It is important here to distinguish algorithmic serendipity from the novum of grace. Algorithmic novelty remains internal to the predictive order: it is variation calibrated against an estimated probability distribution, a more sophisticated extension of the past rather than its interruption. The novum of grace, by contrast, is an ontological break. It is not a less-expected point along a continuum of prior behaviour but the gratuitous irruption of a future that is not derivable from the past, precisely because it is given. Where serendipity refines prediction, grace suspends it. The Counter-Imago can host the former indefinitely; it is structurally inhospitable to the latter.

Bernard Stiegler’s notion of ‘tertiary retention’ is useful here. Where Husserl distinguished primary retention (the living present of perception) from secondary retention (recollected memory), Stiegler identifies a third order: the technical inscription of memory in external media, from writing and print to photographic, cinematic, and now algorithmic supports. Tertiary retention has always co-constituted human thought; every literate culture relies upon it. What distinguishes algorithmic tertiary retention is that it is no longer a passive archive but an active, predictive mechanism that folds stored traces back into present perception as anticipatory salience. Remembrance, in other words, is converted into actionable prediction.Footnote ¹⁷ What is retained is not interior meaning, but behavioural trace. The system ‘remembers’ the user as a pattern of interactions and then projects that pattern forward as a horizon of likely engagement. This creates a functional double of the self, a profile that is treated as the user’s effective identity for the purposes of ranking, recommendation, and monetisation.

I call this the Counter-Imago. The Counter-Imago is not a metaphysical replacement for the imago Dei, nor a literal second self. It is a practical simulacrum, a representation of the person formed from past traces and used to govern what will be shown and reinforced. Its theological significance is that it tends towards closure. It construes the self as what has already been, refined and stabilised into predicted preference, and then returns that stabilised self back to the subject as the world they inhabit. The more such loops dominate, the more the future can feel like a replay rather than a summons.

Here, the difference between data retention and memoria becomes decisive. Pneumatological memory is not only recollection. It is the interior capacity to be addressed, judged, forgiven, and renewed.¹³ It is open to the novum, to what cannot be derived from statistical continuity. Repentance presupposes that the self is not reducible to its past patterns. It presupposes that grace can reorder love and that a person can become other than what their prior history would predict.

The Counter-Imago, by contrast, is structurally inhospitable to repentance. Not because it forbids it, but because it is built to treat the past as the best predictor of the future and to monetise that continuity. The result is a subtle spiritual pressure towards determinism. The self is reflected as a curated repetition of its prior appetites, polished into inevitability. In that sense, algorithmic closure is not merely a distraction. It can become a rival account of identity, one in which the future is primarily an extrapolation, not a promise.

This provides a theological rationale for sanctification as the reordering of desire. If algorithmic habituation tightens the loop of repetition, sanctification must be described as a Spirit-given reopening of time. Charity restores dilatio, not as mere slowness, but as the regained capacity to judge ends truthfully and to desire beyond the constraints of economised salience. In that restored spaciousness, repentance becomes thinkable again, not as self-invention, but as the Spirit’s gift of new love.

This ‘holy hesitation’ should not be read as a generic slowness but as structurally inscribed within the architecture of Thomistic prudence. Aquinas identifies several quasi-integral parts of prudence, among which circumspectio and cautio are especially pertinent here. Circumspectio is the capacity to perceive the wider circumstances in which an act must be judged,Footnote ¹⁸ such that the immediately salient is situated within a broader moral horizon; cautio is the disposition to anticipate and avoid the obstacles that predictable momentum introduces into practical judgement. Together, these parts of prudence provide the structural bridge between the infusion of charity and the concrete resistance to algorithmic acceleration. The Spirit’s dilatio is, thus, not merely an affective slowing but the reactivation of these prudential dispositions, circumspectio widening the field of salience and cautio restoring wariness towards the frictionless, by which grace reopens the practical interval that predictive architectures tend to close.Footnote ¹⁹

The core claim of this article is, therefore, precise: predictive algorithmic systems must be interpreted not only as instruments of action but as ecologies of habituation that shape the preconditions of willing. Two implications follow. First, for theological ethics, an exclusively instrumental framing is incomplete, because it addresses the morality of use while leaving unexamined the formative pressures exerted on attention, memory, and desire. Second, for spiritual theology, the recovery of dilatio emerges as indispensable: the temporal spaciousness in which intellect can judge and will consent without coercive momentum is not a luxury of ascetical refinement but a condition of mature, grace-elevated freedom.

It is important to clarify that ‘the algorithm’ in this argument refers not to a single line of code or an isolated model, but to the integrated architecture of ranking, recommendation, and feedback systems optimised for measurable engagement. These systems operate through probabilistic inference, updating weights in light of past interaction to increase the likelihood of further interaction. Their power lies less in any single prediction than in the iterative loop by which prediction becomes exposure and exposure becomes preference. In that recursive structure, statistical modelling participates in the gradual stabilisation of desire. The theological concern, therefore, is not mystical projection onto machinery, but the cumulative effect of optimisation routines that continually refine what appears before the imagination and thereby condition what becomes thinkable, lovable, and repeatable.

5. Sanctification as the reordering of love

If the algorithmic feed operates on the level of the sensitive appetite, shaping the automatic self before the intellect can engage, then the doctrine of sanctification must be understood as penetrating these same pre-volitional layers. It is insufficient to conceive of holiness merely as a series of conscious ethical choices; rather, it must be understood as an ontological re-scripting of the agent’s interiority. Aquinas’s doctrine of the invisibilis missio Footnote ²⁰ (invisible mission) proves crucial here. In the Summa Theologiae, Thomas teaches that the invisible mission of the Holy Spirit is the divine Person’s indwelling in the rational creature, a mission known not by sensory perception but by its effects: the reordering of desire and the infusion of charity. This mission is not a transient intervention but the establishment of a new mode of existence, inserting a ‘holy hesitation’ into the stimulus-response loop of the algorithmically shaped self. This hesitation creates the ontological space necessary for a freedom that is not merely the selection of pre-fabricated options, but the pursuit of a Good that transcends the digital menu.

This ‘hesitation’ is grounded in the habitus of grace. Aquinas insists that the Spirit’s mission entails the infusion of caritas, which functions not merely as one virtue among others, but as the forma virtutum (the form of the virtues). Charity serves as the gravitational centre that reorders every other habit towards the bonum divinum. The gift of the Spirit is thus a permanent disposition, a religio that restrains the appetite from idolatrous attachment and a timor (holy fear) that anchors the soul in reverent wonder. In the context of the attention economy, the effect of this infused charity is to chasten the accelerated will, slowing the rush to satiation and opening a horizon of dilatio within the soul. This is the Spirit’s modus operandi: reordering created causality without violating it, thereby making room for the soul’s return (redire) to God.

Following the Thomistic moral theologian Servais Pinckaers, we might articulate this conflict as a rivalry between two conceptions of liberty. The algorithm trains the user in ‘freedom of indifference’,Footnote ²¹ the raw capacity to choose arbitrarily among given alternatives (A or B), regardless of their connection to the true Good. The Spirit, conversely, trains the soul in ‘freedom for excellence’, the disciplined capacity to pursue the Good with ease and joy.²¹ Mortification in this context involves the subtraction of those digital phantasms that crowd the imagination and colonise attention, while vivification implants orientations that emerge from the divine economy of desire. Sanctification is thus a pneumatic reordering: the Spirit inscribes the law of the Gospel into the heart, not by destroying the technological mediation of modern life, but by redirecting its pondus towards the summum bonum.

Finally, to guard against a functionalist interpretation, where grace is reduced to a superior mechanism for mental hygiene, we must recall that this ‘re-scripting’ is a participation in the divine nature, not a technique of self-optimisation. The Spirit’s operation is not an efficient cause optimising a system, but a gracious elevation of created agency. Grace does not compete with the algorithmic feed on the same plane of causality; it elevates created agency beyond the closed horizon of predictive formation. The ‘holy hesitation’ is therefore not a product of willpower but a fruit of the Spirit, an interruption of the closed loop of prediction by the sheer gratuity of the divine presence.

6. Eucharistic temporality and the healing of desire

The algorithmic feed operates according to a logic of perpetual dissatisfaction. The Eucharist functions here not merely as a symbolic contrast to digital media but as a counter-habitus in the strict sense. A habitus is a stable disposition formed through repeated acts within a structured environment, and liturgy is precisely such a structured environment: it trains attention through fixed sequences, constrains choice through received forms, and reorders desire by repeatedly presenting the highest good under sacramental signs.Footnote ²² The liturgy does not compete with the feed by offering more compelling stimuli, but by reconfiguring the conditions of salience and judgement. It interrupts the reflex of immediate consumption through practices of waiting, confession, hearing, petition, and thanksgiving, and it culminates in a mode of reception that is not acquisition but communion. In that way, the Eucharist does not merely supply different content; it inculcates a different temporal and affective grammar for what it means to want, to consent, and to rest.

In the Summa Theologiae, Aquinas argues that the human will cannot be satiated by any finite good,Footnote ²³ for the will tends towards the universal good. The feed exploits this structural openness of the will by offering an endless succession of finite particulars, an infinite scroll of images that stimulates the appetite without ever resolving it. This generates a state of inquietudo (restlessness), where the soul is trapped in a cycle of seeking that never finds repose. By contrast, the Eucharist presents the inverse logic: it is a finite quantity (the host) that contains the Infinite Good. Where the algorithm offers a ‘multitude without unity’, scattering attention across a thousand trivialities, the Sacrament offers ‘unity in the fragment’, gathering the scattered self into communion with the Totus Christus.

Temporally, the algorithm collapses the future into a repetition of the past. Because it operates on predictive probability derived from stored data, it can only project a future that resembles what has already occurred. The Eucharistic liturgy, however, breaks this closed temporal circuit. Aquinas teaches that a sacrament is a sign that reminds us of the past (Christ’s Passion), demonstrates the present (grace), and prognosticates the future (glory). The epiclesis, the invocation of the Spirit, is the moment where this threefold time irrupts into the present. It declares that the future is not a calculation based on user metrics, but a gift based on divine promise. In this moment, the Church is trained to inhabit a time shaped by hope (spes) rather than probability.

Consequently, the liturgical act serves as a counter-formation to the ordo extractionis. The ‘Amen’ at the conclusion of the Eucharistic Prayer is not merely a verbal assent; it is an ontological refusal of the ‘Refresh’ command that drives the digital economy. The ‘Refresh’ seeks the next stimulation, driven by the vice of curiositas (wandering curiosity); the ‘Amen’ rests in the truth received, exercising the virtue of studiositas (temperate attention).Footnote ²⁴ By participating in this sign, the community rehearses its identity: not as a data set to be optimised, but as a body to be sanctified. The Eucharist does not merely provide an alternative content to the feed; it provides an alternative form of existence, one in which the self is secured not by predictive analytics, but by the Word and Spirit of God.

7. Ecclesial prudence and fraternal correction

If the preceding analysis is correct, sanctification as the reordering of desire cannot remain a purely interior affair. The Counter-Imago operates through a kind of enforced solitude, in which the user is mirrored back to themselves by a profile refined from their own past. The theological remedy to this closure cannot itself be solitary. It must be ecclesial, enacted through the shared life of the Body of Christ, where the will is formed within a community whose practices resist the smooth feedback of algorithmic reinforcement.

The sources for this claim are Thomistic rather than generically ‘ecclesial’. Aquinas treats prudence not merely as an individual capacity but as perfected within a shared order, distinguishing prudentia politica Footnote ²⁵ precisely to name the virtue’s communal exercise. He also identifies memoria as a quasi-integral part of prudence (Summa Theologiae, II–II, q. 49, a. 1),Footnote ²⁶ such that the truthful recollection of the past is integral to right judgement of what is to be done. He further treats fraternal correction as an act of charity ordered to the spiritual good of the neighbour (Summa Theologiae, II–II, q. 33, a. 1). These three loci, prudentia politica, memoria as part of prudence, and correctio fraterna, Footnote ²⁷ supply the specifically Thomistic grammar for what follows.

Within that grammar, I identify three movements by which the Church, as hermeneutical community, interrupts the habituation of the Counter-Imago. Each corresponds to an operation of practical reason under grace.

The first movement is naming (confessio). The community assists the individual in identifying, with theological precision, the rival mediators at work within the feed. What presents itself as ‘connection’ may in fact be acedia; what appears as harmless curiosity may be curiositas, the restless diffusion of attention against which Aquinas contrasts studiositas. This naming is not moralistic denunciation but the disciplined recollection proper to memoria as part of prudence: it restores the past to truth by refusing the euphemisms through which predictive architectures render their formative pressure invisible.

The second movement is discerning (iudicium). Once named, the passion must be judged by its end. The community evaluates whether the desires cultivated by the feed open the self towards the neighbour and towards God, or whether they collapse the self inward upon the profiled identity. This is the practical exercise of the mind of Christ within the digital environment, distinguishing the bonum honestum from the merely bonum delectabile of the scroll.

The third movement is receiving correction (correctio fraterna). Here, the private ‘truth’ constructed by the algorithmic profile is submitted to the shared discernment of the ecclesial body. Aquinas treats fraternal correction as an act of charity ordered to the healing of spiritual blindness and the restoration of right judgement. In the context of predictive systems, such correction functions as a counter-formative practice by which the Spirit loosens the hold of habituated determinism. The friction of communal love, the resistance offered by embodied, unoptimised persons, interrupts the smooth feedback loops of algorithmic reinforcement.

These three movements are not a programme for technological withdrawal. They describe, rather, how sanctification as the reordering of desire takes ecclesial form under conditions of algorithmic mediation: through truthful speech and faithful presence within, not away from, technologically shaped life. They are offered here as a forward-looking extension of the article’s central argument. A fuller account of their ecclesiological and pastoral implications must await a separate treatment.

8. Conclusion: the teleology of rest and the limits of prediction

The argument advanced here does not entail a rejection of technology as such, nor a Gnostic suspicion of material mediation. The difficulty posed by the algorithmic feed lies not in digital materiality, but in the teleology embedded within its optimisation logic. In his treatment of beatitudo, Aquinas distinguishes between the natural desire for the summum bonum, infinite because God is inexhaustible,Footnote ²⁸ and the restless appetite for finite goods, which appears infinite only because no finite object can satisfy the will’s orientation towards the universal good. The algorithmic feed structurally aligns with the latter dynamic. It generates successive stimulation without terminus, sustaining motion without repose. The result is not fulfilment but perpetual activation, a patterned inquietudo that habituates the will to recurrence rather than consummation.

The Christian response, therefore, is not iconoclastic rejection but teleological discernment. Technologies must be situated within a broader account of the human good. The Church’s witness to the visio beatifica names a future that is promised rather than probabilistically inferred, and this distinction clarifies the anthropological limit of prediction.

Predictive systems can model behavioural regularities with increasing accuracy by learning from observable traces. They forecast tendencies in actus hominis, recurrent patterns of response often enacted with minimal reflective engagement. Moral acts, however, are formally specified by interior intention and consent, not merely by externally describable behaviour. For that reason, predictive accuracy does not exhaust voluntary causality. Statistical inference correlates patterns; it does not penetrate the act insofar as it is freely chosen under the aspect of the good. The limit in question is anthropological rather than computational. As Aquinas argues in his discussion of created intelligences, even superior created minds may infer much from effects, yet the secrets of the heart remain inaccessible except to God and the agent.Footnote ²⁹ In this precise sense, algorithmic prediction encounters a boundary. It may anticipate patterned responses, but it cannot foreclose repentance, because repentance is not an extrapolation of the past but a reorientation of love that opens a genuinely new horizon of action.

Within this eschatological frame, sanctification appears not as the optimisation of a behavioural profile but as the gradual transfiguration of the person through infused charity. The analogy proposed earlier can now be restated with restraint. The feed functions as an extractive economy of attention that tends towards probabilistic continuity; the Spirit’s invisible mission operates as a gratuitous gift that reorders love towards communion. The former stabilises preference within measurable loops; the latter enlarges desire beyond what prior behaviour would predict. The feed reflects the history of consumption; sanctification re-narrates the self in light of promise. The difference is not between technology and grace as rival substances, but between two modes of temporal formation: one oriented towards iterative reinforcement, the other towards participatory beatitude.

It is this difference, articulated within a Thomistic account of habitus and charity, that a theology of sanctification is uniquely positioned to clarify in the age of predictive systems.