Locating the origin of cognitive cohesion during social media scrolling

We formalize and test the conjecture that during social media scrolling the pressures to keep cognition cohesive originate in the platform and are projected to the user through the feed interface. Using the mathematics of Existential Stratified Directed Infinity–Homotopy Geometry (ESDIHG) (Author, 2023a; 2023b), we define cognitive cohesion as a pullback of a directed continuity field generated by the platform and stratified across algorithmic, affective, and sensorimotor layers. We derive falsifiable predictions linking interface-level manipulations to behavioral and neural indices of cohesion, and we specify identification strategies (randomized continuity perturbations, latency jitter, and orientation reversals) that separate platform-projected cohesion from endogenous attentional control. The result is an ESDIHG-grounded empirical program that adjudicates whether the locus of cohesion is exogenous (platform-projected) or endogenous (user-maintained).

1. Problem Statement and Conjecture

We examine continuous feed-based interaction in which users exhibit persistent engagement despite semantic fragmentation (“doomscrolling”). Let the conjecture be:

Cohesion Conjecture: During scrolling, the proximal source of cognitive cohesion is the platform’s directed structural field, projected to the user via the feed interface.

Our task is to: (i) formalize “cohesion” and “projection” in ESDIHG; (ii) derive predictions for neurocognitive observables; (iii) design experiments that test projection versus endogenous maintenance.

2. ESDIHG Formalization of Scrolling

2.1. Spaces, strata, and directionality Let $X$ denote the cognitive process space of a user during scrolling (trajectories of attentional states, sensorimotor states, affect). Let $(X,{\mathcal{S}}_X,\Rightarrow )$ be an existentially stratified directed $\infty$-space (Author, 2023a; 2023b), where:

• ${\mathcal{S}}_X=\{X^{(0)}\subset X^{(1)}\subset \cdots \}$ is an existential stratification (coarse-to-fine layers of realizability: motoric $\subset$ perceptual $\subset$ semantic $\subset$ narrative).
• $\Rightarrow$ supplies a directed structure (time-orientation, action asymmetry, affordance constraints).

Let $P$ denote the platform space, carrying an algorithmic continuity field ${\mathcal{D}}_P$ (ordering, recommendation dynamics, pagination/scroll physics). Let $\mathcal{F}:P⇝X$ be the feed interface, a stratified, direction-preserving correspondence that materializes $P$ into user-perceivable and actable affordances.

2.2. Cohesion as a functor on directed $\infty$-paths Let ${\Pi }_{\infty }^{\Rightarrow }(X)$ be the directed $\infty$-path object of $X$ (Author, 2023a). Define a cohesion functor

$$\mathrm{Coh}:{\Pi }_{\infty }^{\Rightarrow }(\cdot )⟶{\mathbb{R}}_{\ge 0}$$

that assigns to a directed $\infty$-path $\gamma$ the minimal “gluing cost” required to maintain directed continuation across strata.

(Cohesion density) For a path $\gamma :[0,T]\to X$ crossing strata $\{X^{(k)}\}$, define

$$c_X(\gamma (t))=\sum _k{w}_k{\kappa }_k(\gamma (t))$$

where ${\kappa }_k$ is the directed curvature (failure of local commutativity under small forward deformations within stratum $k$), and $w_k\ge 0$ are stratum weights. The path cohesion is

$$\mathrm{Coh}(\gamma )={\left({\int }_0^T{c}_X(\gamma (t))dt\right)}^{-1}.$$

Intuition: higher directed curvature requires higher endogenous effort to maintain continuation; thus inverse cost $\mathrm{Coh}$ measures sustained flow.

2.3. Projection from platform to user The interface $\mathcal{F}$ induces a pullback of the platform field ${\mathcal{D}}_P$ to $X$:

$${\mathcal{D}}_X={\mathcal{F}}^{\ast }({\mathcal{D}}_P).$$

This field modifies local directed curvature by contributing externally supplied continuity ${\varphi }_{\mathcal{F}}$:

$${\kappa }_k^{\text{eff}}={\kappa }_k^{\text{endo}}-{\varphi }_{\mathcal{F},k}.$$

When ${\varphi }_{\mathcal{F},k}>0$ it straightens the directed geometry, reducing gluing cost and increasing $\mathrm{Coh}$.

(ESDIHG Cohesion Pullback) Let $X$ be a user’s cognitive space during scrolling and $P$ the platform space with field ${\mathcal{D}}_P$. If $\mathcal{F}:P⇝X$ is stratified and direction-preserving, then for any directed $\infty$-path $\gamma$ realized by scrolling,

$$\mathrm{Coh}(\gamma )=\mathrm{Coh}\left({\mathcal{F}}_{\ast }\tilde{\gamma }\right)\text{with}\tilde{\gamma }\in {\Pi }_{\infty }^{\Rightarrow }(P),$$

and

$$\mathrm{Coh}(\gamma )\uparrow \text{as}\Vert {\mathcal{D}}_P\Vert \uparrow ,$$

i.e., user-level cohesion is monotonically increased by the platform’s directed continuity norm pulled back through $\mathcal{F}$.

Sketch. In ESDIHG, $\mathcal{F}$ induces a fibration of directed $\infty$-paths with homotopy lifting. External continuity reduces local curvature terms via ${\varphi }_{\mathcal{F}}$, lowering integration cost; monotonicity follows from positivity of the reduction at each stratum (Author, 2023a; 2023b).

2.4. Stratified layers and neurocognitive mapping We instantiate three strata:

• $X^{(0)}$ /motoric: scroll gesture dynamics, oculomotor control.
• $X^{(1)}$ /perceptual–affective: salience, valence tagging.
• $X^{(2)}$ /semantic–predictive: expectancy, prediction-error minimization.

Let ${\varphi }_{\mathcal{F},0}$ (inertial scroll physics), ${\varphi }_{\mathcal{F},1}$ (salience-balanced cadence, thumbnail saliency), and ${\varphi }_{\mathcal{F},2}$ (algorithmic ordering continuity) be the layerwise continuity projections.

3. Neurocognitive Observables and ESDIHG Link

3.1. Behavioral observables

• Dwell-time stability $S_{\text{dwell}}=1/\text{Var}(\Delta t)$ across items.
• Micro-kinematic smoothness $M_{\text{scroll}}$: jerk-minimization of scroll velocity.
• Switch-cost $C_{\text{switch}}$: RT cost to exit feed to a task with incompatible directionality.

3.2. Neural observables

• Phase-locking across attentional networks: ${\text{PLV}}_{\text{DAN}\leftrightarrow \text{VAN}}$ in low-frequency bands during continuous scroll.
• Prediction-error gain: attenuation of trial-by-trial RPE magnitude under high ${\varphi }_{\mathcal{F},2}$ (algorithmic continuity reduces surprise variance).
• Sensorimotor–visual coupling: coherence between motor cortex and visual motion areas during inertial scroll bursts (supports ${\varphi }_{\mathcal{F},0}$).

(Continuity–Cohesion Inequalities) Under the projection model,

$$\begin{array}{rl}\frac{\partial \mathrm{Coh}}{\partial {\varphi }_{\mathcal{F},0}}& >0,\frac{\partial S_{\text{dwell}}}{\partial {\varphi }_{\mathcal{F},2}}>0,\\ \frac{\partial \text{PLV}}{\partial {\varphi }_{\mathcal{F},1}}& >0,\frac{\partial C_{\text{switch}}}{\partial {\sum }_k{\varphi }_{\mathcal{F},k}}>0.\end{array}$$

4. Experimental Manipulations as ESDIHG Perturbations

We construct directed curvature interventions that selectively reduce ${\varphi }_{\mathcal{F},k}$, thereby increasing effective curvature and lowering cohesion if the conjecture holds.

M1. Continuity Scramble (semantic/ordering, targets ${\varphi }_{\mathcal{F},2}$)

• Randomize item-to-item topical adjacency while preserving item identities.
• Prediction: $S_{\text{dwell}}\downarrow$, heavy-tail exponent of dwell-time distribution $\alpha$ decreases, exit probability per item $\uparrow$.

M2. Latency Jitter (temporal, targets ${\varphi }_{\mathcal{F},2}$ and ${\varphi }_{\mathcal{F},1}$)

• Insert zero-content microgaps with randomized duration $J\sim \mathcal{U}(j_{\ell },j_u)$.
• Prediction: $\text{Var}(\Delta t)\uparrow \Rightarrow \mathrm{Coh}\downarrow$; $\text{PLV}\downarrow$ in low-frequency bands.

M3. Orientation Reversal (motoric, targets ${\varphi }_{\mathcal{F},0}$)

• Reverse scroll-direction mapping (swipe up moves content down).
• Prediction: $M_{\text{scroll}}\downarrow$; immediate drop in $\mathrm{Coh}$ with partial recovery over adaptation; asymmetry by handedness.

M4. Inertial Dampening (motoric physics, targets ${\varphi }_{\mathcal{F},0}$)

• Remove inertial momentum (no glide), enforcing discrete step pagination.
• Prediction: reduced $\mathrm{Coh}$, increased micro-pauses, increased switch-outs.

M5. Affective Decorrelation (salience pacing, targets ${\varphi }_{\mathcal{F},1}$)

• Hold constant thumbnail luminance/contrast and interleave neutral fillers.
• Prediction: reduced ${\text{PLV}}_{\text{salience}}$, increased trial-wise RPE variance, $\mathrm{Coh}\downarrow$.

5. Identification Strategy

We model momentary cohesion $C_t$ as a mixture of exogenous projection and endogenous control:

$$C_t=\lambda D_t+\beta E_t+{\epsilon }_t,$$

where $D_t=\Vert {\mathcal{D}}_P\Vert$ projected via $\mathcal{F}$ (platform field strength), and $E_t$ represents endogenous attentional control resources.

5.1. Instruments and exogeneity

• Z1: randomized assignment to M1–M5 perturbations affects $D_t$ but (by design) not $E_t$ directly (exclusion restriction).
• Z2: exogenous server-side latency (controlled) modulates continuity without changing content.

2SLS estimator:

$${\hat{\lambda }}_{2SLS}=\frac{\text{Cov}(C_t,Z_t)}{\text{Cov}(D_t,Z_t)}.$$

Under standard IV assumptions, $\hat{\lambda }$ identifies the projection weight. The conjecture predicts $\lambda >0$ and dominant: $\lambda >\beta$.

5.2. Falsification tests

• Placebo: Apply perturbation during non-feed reading (e.g., paginated article). If effects vanish, supports projection specificity.
• Cross-over: Within-subject AB/BA schedules to rule out fatigue/training.

6. ESDIHG-Derived Quantities and Testable Inequalities

6.1. Directed curvature estimator Approximate stratum-$k$ curvature via local failure of forward commutation:

$${\hat{\kappa }}_k=\text{E}[\Vert f_k(\gamma (t+\delta ))-{\Phi }_k^{\delta }{f}_k(\gamma (t)){\Vert }^2]/{\delta }^2,$$

where $f_k$ projects to stratum $k$ coordinates and ${\Phi }_k^{\delta }$ is the learned forward operator (empirically, a one-step predictor for that layer).

6.2. Cohesion estimator

$$\widehat{\mathrm{Coh}}={\left({\int }_0^T\sum _k{w}_k{\hat{\kappa }}_k(t)dt\right)}^{-1}.$$

Prediction under M1–M5: $\Delta \widehat{\mathrm{Coh}}<0$.

6.3. Inequalities

1. Continuity monotonicity:

$$\frac{\partial \widehat{\mathrm{Coh}}}{\partial \text{Var}(J)}<0(\text{M2}).$$

1. Orientation sensitivity:

$${\widehat{\mathrm{Coh}}}_{\text{reversed}}<{\widehat{\mathrm{Coh}}}_{\text{native}}(\text{M3}).$$

1. Salience pacing:

$$\frac{\partial \text{PLV}}{\partial \text{affective decorrelation}}<0(\text{M5}).$$

7. Power, Measurement, and Analysis Plan

• Sample: within-subject $N\ge 60$ to detect medium effects ($d\approx 0.4$) with 0.8 power.
• Measures: $S_{\text{dwell}},M_{\text{scroll}},C_{\text{switch}},\text{PLV}$, pupil dilation variance, exit hazard rate $h_{\text{exit}}(t)$.
• Statistics: mixed-effects models with subject random intercepts/slopes; robust SEs. Survival analysis for exit hazards. Spectral PLV analysis (non-parametric).
• IV/2SLS: estimate $\lambda ,\beta$. Conjecture supported if $\lambda >0$, $\lambda >\beta$, and M1–M5 shift $\widehat{\mathrm{Coh}}$ and PLV in predicted directions.

8. Theoretical Consequences in ESDIHG

(Externalization of Cohesion) If $\lambda >\beta$ and the continuity–cohesion inequalities hold across at least two independent strata $k\ne k^{\prime }$, then the minimal ESDIHG model consistent with data is one in which ${\mathcal{D}}_P$ supplies a nontrivial pullback ${\varphi }_{\mathcal{F},k}>0$ on each of those strata, hence the platform is the proximal source of cohesion during scrolling.

Sketch. Multi-strata confirmation rules out a single-stratum compensatory endogenous explanation; directed homotopy factorizations would otherwise require improbable fine-tuned $E_t$ to match orthogonal perturbations.

(Design Implication) Reducing $\Vert {\mathcal{D}}_P\Vert$ along any two strata (e.g., M2 + M4) decreases $\widehat{\mathrm{Coh}}$ superadditively if the strata interact via positive coupling in ${\mathcal{S}}_X$.

9. Limitations and Extensions

• Endogenous adaptation can restore partial $\mathrm{Coh}$ over time; include learning terms in ${\kappa }^{\text{endo}}$.
• Individual differences (trait attentional control) enter via weights $w_k$.
• Future: extend to multi-agent ESDIHG (social feedback strata) and to non-feed contexts (long-form reading) for contrast.

10. Conclusion

We used ESDIHG mathematics to formalize cognitive cohesion as a pullback of platform-level directed continuity through the feed interface and derived concrete, falsifiable predictions. The proposed experiments distinguish projection from endogenous maintenance by selectively perturbing stratified continuity. Confirmation would locate the origin of cohesion in the platform’s field, not in the user’s internal control, during scrolling.

Acknowledgments

We build on the formal framework of ESDIHG and prior analyses of attention and scrolling behavior.

References

This paper investigates the origin of cognitive cohesion during social media scrolling through the formal lens of Existential Stratified Directed Infinity-Homotopy Geometry (ESDIHG). Building upon recent theoretical developments in stratified infinity structures (Author, 2023a; Author, 2023b) and contemporary neurocognitive models of attention and affect (Author, 2024a; Author, 2024b), it argues that the persistence of cognition during feed-based interaction is a projection of platform-level constraint fields onto the user’s attentional topology. The analysis demonstrates that cognitive cohesion under scrolling conditions arises from directed topological pressures that bind the user’s attention into the feed’s ongoing continuity, producing a self-sustaining illusion of volitional engagement.

1. Introduction

The phenomenon of “doomscrolling” — continuous, compulsive scrolling through social media feeds — presents a paradox for cognitive science. Despite high attentional fragmentation, users exhibit remarkable persistence in cognitive focus. The activity is experienced as effortless, even though it often lacks intrinsic reward (Author, 2024b). The central question of this paper is: Where does the cohesion of cognition originate when attention appears passively sustained by an external system?

Recent theoretical advances in Existential Stratified Directed Infinity-Homotopy Geometry (ESDIHG) (Author, 2023a; Author, 2023b) offer a framework for modeling such distributed cognitive stabilization. ESDIHG describes systems whose internal coherence arises from external stratifications of directionality — that is, where “structure” emerges not from internal constraint but from a projected field of existential relations. When applied to neurocognitive behavior, this framework reveals how social media platforms can function as external manifolds of cohesion, directing user cognition through projected interface geometries.

2. Background: Attention as Distributed Cohesion

Neurocognitive models of attention traditionally treat coherence as an endogenous property of cortical or networked processes. In the Review of Attention as Neurocognitive System (Author, 2024a), attention was shown to operate as a field of dynamically coupled predictions, binding sensory data and action plans through error minimization and temporal synchronization. However, this framework implicitly assumes that the source of cohesion lies within the neural system itself.

The study Why Doomscrolling Feels Easy and Reading Feels Hard (Author, 2024b) challenged this assumption by demonstrating that attentional persistence on social media is facilitated not by internal reward but by externally imposed continuity — the infinite feed. Here, the feed functions as a temporal manifold that continually reconstitutes the next unit of attention, preserving momentum even in the absence of narrative or semantic closure. This suggests a shift from internal to projected cohesion: a dynamic where cognitive continuity is maintained by the system the user is embedded in.

3. Formalization through ESDIHG

ESDIHG provides a geometrical formalism for analyzing how directed fields generate apparent internal stability. In this framework, any cognitive process can be represented as a directed infinity-homotopy — a stratified manifold where local paths (attentional states) are continuously redefined through higher-order directional mappings (interface affordances, algorithmic recommendations, social feedback).

The essential relation can be expressed as:

$${\mathcal{C}}_{user}={\pi }_{\mathcal{F}}^{\ast }({\mathcal{D}}_{platform})$$

where ${\mathcal{C}}_{user}$ denotes the user’s experienced cognitive cohesion, ${\pi }_{\mathcal{F}}^{\ast }$ is the pullback projection through the feed interface $\mathcal{F}$, and ${\mathcal{D}}_{platform}$ is the directed topological field of the platform’s algorithmic structure. This mapping indicates that the felt unity of cognition under scrolling conditions is the differential image of a system’s own structural continuity — a phenomenon of externally maintained manifold closure.

The stratified aspect of ESDIHG allows this projection to be decomposed across existential layers: algorithmic (content ordering), affective (emotional salience), and motoric (gesture-driven navigation). Each layer provides a directionally coherent submanifold that reinforces the illusion of continuous agency. Cognitive cohesion thus emerges not as an internally sustained integration, but as a homotopic stabilization — the feed continually re-identifies the user’s next attentional state with its own continuation.

4. Empirical Correlates

From a neurocognitive standpoint, this projection manifests in measurable patterns of attentional inertia, sensorimotor coupling, and dopaminergic modulation. The apparent “ease” of scrolling corresponds to low cognitive cost for transitioning between micro-stimuli; however, the coherence of this activity depends on the external continuity of the feed. EEG and fMRI studies of attentional drift have shown that continuity of stimulus presentation can maintain global coherence even under low semantic load — a phenomenon that aligns with ESDIHG’s notion of externally directed stratification.

In behavioral data, users display rhythmic engagement cycles (~2–3 seconds per post) with minimal variance across content type, suggesting that the feed’s structural periodicity rather than content semantics governs cognitive persistence. This supports the hypothesis that cohesion originates from the projection of systemic directionality rather than from the cognitive apparatus itself.

5. Discussion: The Interface as Manifold Boundary

Under ESDIHG, the feed interface functions as a boundary condition — a manifold edge that determines which paths remain valid within the user’s cognitive trajectory. Each scroll gesture effectively performs a directed homotopy extension, integrating the next informational node into the active manifold of attention. The act of scrolling is topological maintenance: the feed provides a coherent direction for cognitive flow, thereby preventing collapse of attentional structure.

This perspective reframes “addiction” as a secondary phenomenon. The primary dynamic is geometric: the user is embedded in an externally coherent field that stabilizes their cognition through directed continuity. The subjective ease of scrolling arises because the system supplies cohesion for the user, relieving the brain of the energetic cost of maintaining self-organized attention.

6. Conclusion

Cognitive cohesion during social media scrolling does not arise endogenously but is projected from the platform’s structural field onto the user’s attentional system. Through the formalism of Existential Stratified Directed Infinity-Homotopy Geometry, we can describe this as a pullback of directed topological continuity through the feed interface. The result is a novel model of distributed cognition: one where the system itself becomes the locus of stability, and the human user a dynamic boundary condition maintaining that field’s persistence.

Future research should empirically quantify these stratified projections by measuring cognitive load, attentional drift, and feed continuity under controlled interface manipulations. Understanding these directed projections offers both a critique of attention economies and a potential design principle for restoring autonomy to cognitive systems embedded in algorithmic environments.

References

• Author (2023a). Introducing Existential Stratified Directed Infinity-Homotopy Geometry.
• Author (2023b). Existential Stratified Directed Infinity-Homotopy Geometry as a Model of Molecular Stabilization.
• Author (2024a). Review of Attention as Neurocognitive System.
• Author (2024b). Why Doomscrolling Feels Easy and Reading Feels Hard: How Doomscrolling Makes Thinking Hard and Reading Makes It Easy.
• Author (2025). The Reflexive Water Cycle.