What it all means.

1.1The cohort-wide flow is the load-handling signature elite cognition produces under sustained competitive pressure. Before adversarial pressure is present (pre-round and early mid-round) elite participants show a varied, expansive attentional flow; during a fight that flow collapses as attention zeroes in on the kill; once the fight resolves, verbalisation swells back up during reflective post-fight and post-round windows. The during-fight collapse reads cleanly as procedural automaticity displacing verbal reporting, since time-pressured execution cannot tolerate the dual-task load of explicit verbalisation alongside motor and perceptual demand (Beilock & Carr, 2001; MacIntyre et al., 2014). The thinning of speech during the fight reflects what cognition can output under that load; even spontaneous combat verbalisation is pared to terse callouts. Both reinvestment theory (Masters, 1992) and the integrationist account of automatic-and-reflective expert processing predict this kind of phase-locked alternation, and the round-level scenarios that follow show how the same flow shifts when round conditions or outcomes change.

1.2Attentional flow appears to respond differently to round conditions and round outcomes. The three round-level scenarios show the same flow with variable allocation of capacity. High-performance won rounds show forward-loaded planning, with pre-round and pre-fight cognitive volume rising against the cohort baseline, carried by Attentional Control and Game Situation rather than by Attentional Metacognitions. This is consistent with the externalised-attention literature, and reframes Lamers and O'Connor's (2023) null result for attentional-focus instructions in FPS aiming as a methodological-design issue. The cohort's data describe FPS attention as a cycling flow across the round-fight cycle, not the stable single-direction anchor that Wulf-style instructions assume. An adversarial, non-self-paced environment makes any single attentional anchor difficult to sustain and likely counter-productive: elite play depends on continual re-routing of attention between teammates, opponents, ability cooldowns, and map state, with the optimal direction shifting cell-by-cell. Davies et al. (2014) note that the post-shot routine literature is dominated by structured procedures for poor shots, with no equivalent routine described for successful shots; the cohort's HP-won pattern instantiates that asymmetry, treating success as requiring less explicit post-event processing than error. Forward-loading is one structural marker of effective execution in this cohort, but the high-performance data also show on-the-fly tactical adaptation: what players term "mid-rounding", where the team re-plans live as the round state shifts away from the original call. Effective execution sits in the combination of forward-loaded planning and mid-round flexibility.

1.3Low-performance lost rounds invert the pattern. Mid-round-post-fight Attentional Metacognition density rises more than threefold above the high-performance figure and roughly 1.6× the cohort baseline. Technical feedback and metacognitive appraisal drive the spike, consistent with reinvestment theory (Masters, 1992): performance error under elite-tier pressure precipitates a return to conscious internal monitoring. Calmeiro et al. (2010) reported the same temporal contingency in skilled golf, with negative appraisals clustering immediately before and after missed targets. An open question this dataset cannot answer is whether prior metacognitive critique then drives more conscious awareness of automatic skills in subsequent rounds, increasing the likelihood of further failure; testing that pattern would require a larger sample of metacognitive autopsy followed by anticipation and present critique. The LP-lost signature, in short, is the inverse of the HP-won one: extra cognition is allocated backward rather than forward.

1.4Eco rounds redistribute the flow along a different axis. Where economy forecloses the normal execution route, Game Situation share rises at the no-combat windows; elite players visibly redirect cognitive resource toward state-monitoring once the play book is constrained. The flow shifts from execution cognition to information-gathering cognition.

2.1Three attentional strategies travel cleanly across self-paced, endurance, and team-tactical sports. The first is the externalisation of expert attention. Across golf (Oliver et al., 2021), cycling (Whitehead et al., 2018), and tennis (McPherson & Kernodle, 2003), expert attention weighs toward environmental and tactical cues over internal mechanics, paralleling the external-focus literature in motor learning (Wulf, 2013). The Valorant cohort replicates this signature more strongly still, with AC-External outweighing AC-Internal at every temporal cell: the cohort ratio is 4.4:1.

2.2The second pattern is the dominance of planning and post-event diagnosis. Calmeiro and Tenenbaum (2011) reported that experienced golfers spent more time on assessing conditions and planning prior to the putt and produced more diagnostic verbalisations after it, often chaining post-shot diagnosis into planning for the next attempt. Whitehead et al.'s (2016) skill-by-pressure data show high-skill golfers' baseline cognition dominated by planning, information-gathering, and club selection. The seven-cell flow on this site maps onto the same load distribution, with pre-round and pre-fight functioning as AC-and-GS-heavy planning windows and mid-round-post-fight as the densest post-event diagnostic cell. Planning-plus-diagnosis dominance therefore generalises from self-paced sport into team-tactical FPS performance with the same seven-cell flow intact.

2.3The third is phase-structured organisation. Whitehead and Jackman's (2021) golf framework, Whitehead et al.'s (2018) cycling time-trial work, and McPherson and Kernodle's (2003) point-by-point tennis representations all describe expert cognition as recursive phase-dependent shifts rather than as a sequential pipeline. The seven-cell crossing of round phase and combat phase is this study's structural extension to a context where each round embeds a complete pre-, during-, and post-fight cycle. The recursive phase-based logic is the prior literature's; the seven-phase crossing is the present contribution.

3.1Four divergences from the prior sport TA work surface in this cohort and look characteristically esports. The first is the richer external-AC content. Where golf TA captures wind, lie, and pin (Oliver et al., 2021), Valorant TA captures enemy position, enemy state, enemy mental modelling, ability cooldowns, and minimap reads. The external object of attention is an adversary rather than a static physical environment, and the codebook inflates accordingly. Where self-paced sports differ from FPS is in what attention attends to; the underlying operation of attention itself stays consistent across both.

3.2The second divergence is the resource-management cognition. Loadout management, economy-aware buying, and eco-round state-monitoring are all live cognitive moves in the dataset. Individual-sport TA has no analogue for these. The closest is perhaps cycling pacing decisions (Whitehead et al., 2018), but even there the resource is a personal energy budget rather than a shared team economy with strategic implications across rounds.

3.3The third divergence is the distributed team cognition. Codes like communication management, collaborative planning, and team-aware planning capture cognition that crosses the boundary of a single player's mind. P01 and P04 demonstrated the active side most clearly, engaging in explicit communication-management strategies that shaped the cohort's collective attentional state from the outside in. P02's session showed the receiving end of the same dynamic: a recorded teammate handled tactical direction, and P02's AC-Internal share dropped below the cohort mean while his Game Situation share rose to the highest in the sample. The cognitive load was real on both sides, just mediated through someone else.

3.4The fourth divergence is the reading of opponents' mental states. Codes capturing mentalising (anticipating an opponent's emotional or cognitive state, then exploiting it for tactical advantage) appear nowhere in the prior sport TA literature, at least not as far as the present author is aware. Self-paced sports have no analogue: the opponent (the course, the wind, the metabolic clock) does not have a mind to read. Elite Valorant players actively model the opposing player's likely behaviour as a basis for anticipation, treating the opposing mind as a tractable object of attention.

4.1The within-elite typology presented in the gallery reads cleanly through Stanovich's (2011) tripartite mind framework, with three distinct signatures observable. Autonomous-mind dominance (P02) appears as below-cohort AC-Internal paired with the cohort's highest Game Situation share, indicating autonomous execution enabled by team-mediated tactical offloading; engagement remains intact even as internal-process verbalisation drops. Algorithmic-mind dominance (P03 and P04) appears as heavy explicit reasoning and theory-of-mind content carrying the cognitive load, with structured tactical models doing more work than experience recall. Reflective-mind dominance (P06) appears as elevated AM with the IGL absorbing team-coordination cognition on top of an individual combat load.

4.2These signatures appear to track structural role (in-game leader, supporting player, lurker), independent of skill hierarchy. Both Radiant- and Immortal-tier participants are represented across the three styles. The pattern warrants greater investigation before it is treated as a stable trait: a defensible typology requires multi-session, multi-role follow-up to distinguish stable cognitive-style traits from session-specific performance-state artefacts. The novice-cohort extension currently in data analysis is the next step toward that disambiguation, and the planned autonomous TA tool may be able to address it further (participants can fulfil every objective point of variance).

5.1Variable Think Aloud reactivity at the elite tier is the most prominent TA-specific limitation in this cohort. Three Radiant-tier participants offered conflicting self-reports: P05 reported no interference; P02 spontaneously remarked that thinking aloud was eliminating his instinct mid-match (a reinvestment-style report consistent with Whitehead et al., 2016); P08 reported the inverse, telling the researcher post-match that thinking aloud helped him stay engaged, a self-regulation facilitation theme also reported by Birch et al. (2022). The within-cohort divergence at the same competitive tier indicates the reactivity question is participant-dependent rather than task-dependent in elite Valorant.

5.2Three further limitations bound the inferences this dataset can support. Single-session capture per participant means cognitive-style signatures cannot yet be disentangled from performance-state fluctuation across matches. Ranked-match capture sits below the pressure ceiling that high-stakes tournament play would impose; reinvestment effects in particular would likely run stronger in tournament conditions, though the present results still map onto the conditions of most players' experience. The inferential design is intentionally descriptive: z-scores indicate cohort magnitude, and no inferential cognitive-style tests have been applied to the underpowered n = 8. Together, these constraints frame the present study as architectural mapping rather than confirmatory testing; patterns are identifiable, but their stability and generalisation remain open empirical questions.

5.3The novice-cohort follow-up (currently in data analysis, first results expected mid-May 2026) is the next step toward elite-vs-novice cognitive comparison on the same coding framework. If you'd like to contribute, register interest in the next study →

6.1If you reference this work in a piece of media, please cite the underlying study:

Langfield, D. C. M. (2026). Meta-attentional flow in elite Valorant
esports: A phase-structured think aloud investigation (Currently
unpublished BSc Empirical Project, Glasgow Caledonian University).

6.2The full empirical report (methods, results, complete discussion, all citations referenced above) is available on request. If the report is later adapted and published in an academic journal, the citation above will be updated to point to the published version. Request the full report