Introduction
Living systems persist by adjusting what they do as their surroundings change. Across taxa, such adjustments often occur in the absence of direct physical force: an organism alters its position, activity, or interaction pattern because some aspect of the situation has become relevant for energy acquisition, safety, or reproduction. 
Despite the centrality of these adjustments to survival, comparative biology lacks a shared, operational way to describe which kinds of informational situations reliably trigger behavior change across species. Most existing approaches focus either on internal mechanisms (neural, physiological, or genetic) or on broad behavior labels that obscure when and why a meaningful switch actually occurs.
Here we take a deliberately external perspective. Rather than attempting to infer internal representations or mechanisms, we treat behavior change itself as the observable footprint of information use and ask a simple comparative question: what kinds of informational situations repeatedly force organisms to change what they are doing?
Every organism faces innumerable tasks in the wild. These tasks are too numerous to catalog exhaustively, and they vary widely across habitats and life histories. Even so, they can be usefully sorted into three broad classes. Physical tasks involve direct forces and contact, such as impacts, crushing, abrasion, or mechanical constraint. Biological tasks concern built-in morphology and chemistry, including digestion, thermoregulation, and baseline metabolic processes. A third class, which is the focus of this study, comprises informational tasks: situations without direct physical impact in which behavior appears to depend on how a configuration in the surroundings bears on the organism’s current requirements, and on selecting among alternative courses of action.
Examples include distant motion, changes in exposure, signals, or shifts in group activity.
Previewing where this analysis leads, our survey identifies five recurrent domains of informational relevance, with five corresponding adaptive tasks: (1) environmental states and fragments, (2) free-moving entities at a distance, (3) context and perception shaping, (4) group dynamics and coalitions, and (5) formal symbolic systems. These tasks exhibit a striking regularity: across the dataset, they partition living organisms into five groups, distinguished by the number of tasks they routinely control—from one to five. Higher-level groups consistently embed and reuse the capacities of lower-level groups while adding one additional task into regular use. This pattern appears across lineages regardless of morphology, habitat, brain size, or even possession of a nervous system.
We present this five-task framing as a working hypothesis rather than a mechanistic theory. The aim is not to close debate, but to offer a compact, testable descriptive scaffold that makes information-sensitive behavior change visible across species.
The remainder of the paper explains how these tasks were accessed observationally; outlines the inclusion and exclusion rules used to identify informational behavior change; introduces the dataset of 1,530 species and its coding protocol [The Five Task Model: From Cognition And Evolution To AGI. DOI: https://doi.org/10.17605/OSF.IO/VB2NC ]; and presents the resulting task structure. We then examine what this five-part framing suggests about information use across life, including why five tasks appear sufficient in this material. Finally, we discuss how the resulting Five Task Model may serve as a reference point for future work in comparative biology, cognitive taxonomy, and the study of artificial systems.

Non-Biological Approach: Information vs. Force

From objects to informational relevance: measuring information use
Field biology provides lavish detail about what organisms touch, build, and endure, but far less about how particular situations become behaviorally relevant at a given moment. Two outwardly similar scenes can drive the same organism toward opposite actions depending on which aspects of the situation are informationally decisive under current conditions. A rustle may signal wind, prey, or a predator; a clearing may function as a basking site or as dangerous exposure. Simply cataloging objects or stimuli does not explain which way a behavior switch will occur. The gap addressed here concerns not objects themselves, but how configurations in the surroundings acquire control relevance in context.
Classic work in ethology and ecological perception already pointed in this direction. Tinbergen and Lorenz emphasized that so-called innate actions are released only under specific configurations rather than by stimuli in isolation, while Gibson’s notion of affordances treated environmental features as relational and context-dependent rather than fixed triggers. In each case, the operative unit was not the stimulus-as-object but the stimulus-as-situation.
Because nearly all habitable environments are unstable and multi-factorial, organisms must continuously adjust to changes in temperature and light, threats and opportunities, distance and resources, and other recurrent perturbations. This requirement appears to hold across life, from bacteria to humans, fungi to insects, plants to primates. Persistence depends not on registering everything, but on selectively responding to those changes that matter under constraint.
As Norbert Wiener observed, “any organism is held together in this action [maintenance of homeostasis] by the possession of means for the accession, use, retention, and transmission of information” (Wiener, 1948/1961). Notably, this formulation does not presuppose neurons, brains, or conscious deliberation. It identifies information—not force—as the organizing constraint.
The means available for such regulation are surprisingly limited. In broad terms, organisms act by adjusting localization (where and how they are situated) and/or interaction (how they engage with other organisms or with internal life-cycle processes such as reproduction). Both forms of adjustment are guided by information about the situation rather than by direct physical contact alone.
Information use in living systems is therefore neither random nor indiscriminate. Even highly stereotyped or “innate” patterns are conditionally released, suppressed, delayed, or redirected depending on contextual cues. Tinbergen’s analyses of releasing mechanisms already made this explicit: particular configurations elicit action when—and only when—specific conditions are met. The relevant unit is not stimulus energy per se, but the situational configuration to which action is selectively coupled.
What emerges is not a picture of blind reflex chains, but of context-conditioned deployment. The capacity to respond differently to the same physical inputs under different circumstances appears less like an optional sophistication and more like a minimal requirement for persistence. Lineages that lack such flexibility, or exhibit it only within a very narrow range, lose the ability to keep environmental change (Point A) aligned with the standing requirements of Energy, Safety, and Reproduction (Point B), and over evolutionary time tend to disappear.
Importantly, many of the informational situations that exert such control are not only species-typical and recurrent, but shared across lineages with radically different morphologies and ecologies. These include, for example:

• Orientation to setting: registering gradients or fragments of the environment and adjusting posture, position, or exposure.
• Interaction with other organisms: engaging with prey, predators, mates, or offspring.
• Regulation of perception: altering visibility, signalling, concealment, or display.
• Life in groups: tracking and responding to group structure, coordination, and competition.

Although diverse in form, these situations converge on a small set of recurring informational tasks. In this article, we focus selectively on such tasks: cases in which behavior change is not driven by direct physical force, but by how situational configurations constrain viable action. By treating tasks at the level of informational relevance rather than contact or object identity, we can compare very different species by the kinds of problems that force behavior change, rather than by the particular mechanisms or materials through which those problems are encountered.

Goals vs. Tasks: the background logic of choice
Before turning to the dataset and the Five Tasks Model, it helps to separate goals from tasks—a distinction that is often blurred in behavioral science. Darwin and later ethologists gestured at it, but it was rarely operationalized. We need it here, because without it we cannot say clearly what counts as an informational task and what belongs to physiology or mechanics.
For living systems, survival appears to rest on three standing conditions: Energy (acquire and conserve resources), Safety (avoid threats and irreversible damage), and Reproduction (secure transfer across generations). Together these form the ESR triad—the permanent goals toward which all adaptive strategies point.
Goals are constants, tasks are the situational problems that arise on the way to them: “Is this patch safe?”, “Approach or avoid that agent?”, “Reveal or conceal now?”, “Coordinate or go alone?”. The ESR triad sets the horizon; tasks are the stepping stones. A lion subduing a zebra is a task serving Energy; a zebra detecting the lion’s approach is a task serving Safety; a bird displaying or building a nest is a task serving Reproduction. Tasks can be brief or extended, novel or routine, but they are all bricks in the same wall.
Put simply: tasks are immediate challenges; goals are enduring requirements. Failure at the level of ESR ends the lineage. Failure at the level of a task is local and episodic—some individuals fail, but the species persists. This is the level at which we can study informational tasks.

Key premise of this work
The central premise of this study is that organisms are both selective and constrained in what they must and can detect and resolve in order to maintain Energy, Safety, and Reproduction (ESR). They do not register and react to everything in their surroundings; instead, they respond to those aspects of the environment that reliably make a difference under prevailing conditions. For this reason, we classify species and their adaptive challenges not by physical objects themselves, but by domains of informational relevance—patterns, relations, or signals that systematically influence behavior. We refer to this background stream as the general informational flow (GIF).
Operationally, this entails shifting the unit of analysis from cataloging environmental items to identifying which environmental configurations reliably precede behavior change across organisms. Framed this way, very different species can be compared by the kinds of informational situations they must and can resolve, even when their bodies, habitats, and life histories are otherwise incomparable.
We are not asking whether a behavior is “adaptive,” “rational,” “cognitive,” or “goal-directed,” nor are we attempting to infer internal mechanisms or representations. The focus is deliberately narrower and external: informational tasks, defined as situations in which action depends on context-sensitive differentiation among alternatives, and in which more than one viable course of action is realistically available to the organism.
This stance is broadly compatible with earlier traditions that treated adaptive action as contingent on situational differentiation rather than fixed stimulus-response coupling, including works by Alexey Leontiev (1948, 1972) on the phylogenetic expansion of environmental differentiation (“Evolution of Psyche”), as well as later comparative approaches that emphasize information use over morphology. 
This survey also resonates with several earlier and contemporary traditions that emphasized the situational use of information in guiding behavior, rather than treating action as a direct function of morphology or stimulus energy alone. These include classic ethological work on context-sensitive releasing conditions (Tinbergen), ecological approaches to perception and affordances (Gibson), comparative analyses of signalling and audience effects (Zahavi; Maynard Smith & Harper), and studies of social coordination and communication in animals (e.g., Cheney & Seyfarth; de Waal; Tomasello).
We do not adopt these frameworks as theoretical foundations, nor do we attempt to unify them. Our contribution is deliberately narrower and more external: we map where behavior actually switches across species and ask whether those switches cluster into a small, ordered set of informational tasks, independent of internal mechanism or cognitive interpretation.

Behavior change as the unit of analysis

Transition from Behavior One to Behavior Two
In this study, the basic observational unit is behavior change (BC) rather than behavior per se. We define a behavior change as a time-locked transition from one observable activity pattern (B1) to another (B2) that occurs in a dynamic context and is plausibly attributable to a non-contact environmental event or configuration.
Methodological illustration: why switches are the handle
If one says that safe driving is simply a way to reach a destination, that is 100% true. An equally true perspective is that safe driving is a sequence of controlled changes in a car’s behavior—timely, appropriate, and precise. We call driving “safe” when the necessary switches occur when they are required: road signs are registered and reflected in changes of speed or direction; other drivers’ maneuvers are met with adequate adjustments; unexpected situations are handled by rapid, appropriate transitions rather than by a fixed continuation.
Now consider an observer who cannot read road signs and does not know traffic rules, but who can reliably detect when a car’s behavior changes. For such an observer, the meaning of a sign is not given in advance. It becomes visible only through systematic associations between particular roadside configurations and subsequent behavior changes—slowing, stopping, turning, yielding. The signs do not physically move the car; their relevance is inferred from the switches they reliably elicit.
The same logic applies in the wild. If we want a comparative, substrate-neutral handle on what matters to an organism—without assuming shared symbols, perceptual categories, or internal representations—we can track which situational configurations reliably precede switches from one ongoing pattern (B1) to another (B2). Continuous activity without switching, however complex, offers little leverage; the informative events are the transitions.
For the purposes of this study, a behavioral state is defined by the organism’s current mode of activity and its salient parameters (e.g., direction, speed, orientation, interaction pattern, or engagement status). A behavior change therefore includes any detectable shift in these parameters, regardless of magnitude, provided it meets the inclusion criteria specified below.
This choice is methodological rather than theoretical. Labels such as foraging, vigilance, or display vary widely across taxa and observers and often obscure whether any information-sensitive adjustment is occurring at a given moment. By contrast, a transition from one ongoing pattern to another (B1→B2) is a discrete, countable event that can be temporally aligned with changes in the surrounding informational flow.
From this perspective, a system counts as living and adaptive not because of its biological substrate, but because it can change its behavior in ways that keep environmental change aligned with Energy, Safety, and Reproduction.
Throughout the paper, the terms behavior change and controlled behavior change are used interchangeably. By controlled we do not mean consciously deliberated, goal-reflective, or cognitively explicit. We mean only that, at the moment of transition, more than one continuation was realistically available, and that the continuation observed was consistent with maintaining energy, safety, or reproduction (ESR) under the prevailing conditions.

Behavior change — localization and interaction*
From an observational standpoint, behavior change becomes visible through two broad classes of adjustment: changes in localization and changes in interaction. Localization includes shifts in position, orientation, exposure, or spatial relation to features of the environment.

*Interaction includes the initiation, modulation, or suppression of engagement with other organisms or with life-cycle processes expressed through observable engagement (e.g., reproduction, provisioning, care).

Importantly, these two forms are not mutually exclusive and often co-occur. A single behavior change may involve relocation without interaction, interaction without overt movement, or coordinated adjustments of both. The distinction is introduced not as a theoretical claim about mechanisms, but as an observational guide: across taxa, survival-relevant behavior change is reliably expressed as a change in where the organism is situated and/or how it engages with relevant entities or processes. This framing allows behavior change to be identified even when locomotion is minimal, constrained, or stereotyped, by tracking shifts in interaction targets, timing, or engagement state.

Why behavior change is used instead of “response” or “reaction”
The terms response and reaction are intentionally avoided as primary units of analysis because they do not require a transition, do not specify alternatives, and are routinely applied to non-living systems. A response may occur without any change in ongoing activity, and a reaction may describe a mechanically coupled effect with no selectable alternatives.
By contrast, behavior change specifies:

• a discrete transition (B1→B2),
• a temporal anchor (the onset of the switch),
• a choice structure (more than one viable continuation),
• and a functional constraint (consistency with maintaining Energy, Safety, or Reproduction).

This framing allows information use to be studied across radically different organisms without assumptions about internal mechanisms, representational formats, or conscious deliberation. The approach is substrate-neutral but empirically constrained: if a putative informational cue is removed, masked, or rendered ambiguous, the corresponding behavior changes should diminish, delay, or diversify.
Many of the transitions captured by this definition occur in domains historically discussed under the heading of instinct. However, classical ethology already treated instinctive behavior not as a blind reflex, but as selective and conditional. Darwin, Tinbergen, and later ethologists repeatedly emphasized that such actions are sensitive to context, timing, and competing pressures, even when they are rapid or stereotyped in form. The present framework does not revive instinct as a theoretical construct; it simply adopts the same observational stance toward selectivity and conditionality.
As an illustrative analogy, behavior change can be thought of as a windsock for information use. A visible switch indicates that something in the surroundings changed in a way that mattered for maintaining ESR, and that this change was followed by a different movement or interaction. The exact content or internal encoding of that information need not be recovered for the transition itself to be registered.
At the same time, we do not claim that every behavioral switch constitutes information use. The claim is narrower and asymmetric: whenever Energy, Safety, or Reproduction are genuinely at stake in a dynamic context, effective control appears to require an appropriate behavior change, typically via a shift in localization, interaction, or reproductive timing.
In this sense, behavior change functions as an observable footprint of upstream information use. It does not specify how information is processed, only when its use becomes behaviorally consequential.

From behavior change to research questions
Because BC can be observed, registered, and counted without cataloging internal mechanisms or arguing over behavior labels, it becomes a good comparative handle. From this vantage, three questions follow quite naturally:

1. Which kinds of cues, signals, and situations in the informational flow reliably trigger species-typical (species-wide) behavior change?
2. Which of these informational events and tasks must (and can) organisms recognize and process in order to maintain ESR triad?
3. Which of these are the most basic and widely shared—common across all or major groups of species?

Answering these links three levels in a single, operational chain: behavior change → informational domains → informational adaptive tasks. In the dataset we count BCs that are (a) triggered by non-contact informational events and (b) offer at least two viable continuations, so that a selection is plausible. Crucially, the meaningful event has to be singled out—isolated from the ongoing informational flow—before a switch can be made. In this survey, 1,530 species are summarized by their capacity and regular need to address such tasks via behavior change.

Inclusion criteria: when a transition is treated as informational
Not all behavioral transitions are treated as evidence of information use. To minimize interpretive overreach, we adopt conservative inclusion criteria. A behavior change is treated as informational only when all three conditions below are satisfied.

1. Non-contact trigger
2. The event plausibly precipitating the transition is not a direct physical impact (e.g., no bite, strike, collision, or mechanical forcing). The putative trigger is distal, configurational, or contextual (e.g., motion at a distance, a change in exposure, a signal, a group shift).
3. Multi-valued situation
4. The situation supports at least two biologically credible interpretations or continuations for that lineage in that context (e.g., approach vs. avoid; reveal vs. conceal; recruit vs. abstain).
5. Situations that reliably and invariantly produce a single motor outcome across contexts are excluded.
6. Selectable alternatives
7. The observed transition (B1→B2) could plausibly have been otherwise under similar conditions and is documented as such in closely related species, neighboring populations, or different contexts within the same lineage. This criterion is crucial: it distinguishes transitions consistent with selection among viable alternatives from fixed closed-loop reactions.

Only when all three conditions are met do we treat the transition as an informationally mediated switch.
In this study, a “viable alternative” does not refer to an exhaustive set of possible actions or to internally represented options. It denotes a small set of ecologically plausible continuations of ongoing behavior that the organism is physically capable of executing in the given context (e.g., approach or withdraw; remain or relocate; display or conceal; recruit or abstain). A behavior change (B1→B2) is treated as selection among alternatives when at least two such continuations are credible for that situation and when different outcomes are observed across contexts, individuals, or closely related cases. This criterion distinguishes controlled behavior change from invariant coupling, where the same environmental trigger reliably produces the same adjustment regardless of context. No assumptions are made about deliberation, optimality, or internal representation; the requirement is solely that behavior shows context-dependent modulation consistent with maintaining Energy, Safety, or Reproduction.

Exclusion criteria and boundary cases
Several classes of phenomena are explicitly excluded from being treated as informational tasks in this study.
First, direct-impact loops are excluded.
Mechanical forcing, chemical shocks, or other forms of direct physical impact that reliably produce the same behavioral outcome regardless of context are not treated as informational behavior changes. In such cases, no selection among alternatives is observable at the behavioral level.
Second, always-on morphological or biochemical traits are excluded.
Features such as aposematic coloration, structural spines, constitutive toxin production, or constant luminescence are treated as biological carriers rather than instances of information use, unless there is clear evidence that their expression is selectively adjusted in real time (e.g., shown vs. concealed, activated vs. suppressed) in response to changing conditions.
Third, invariant couplings between environmental change and behavioral adjustment are excluded. Behavioral transitions are not treated as informational when a given class of environmental change reliably produces the same adjustment across contexts, with no observable branching into alternative continuations. In such cases, although behavior may change over time, there is no evidence at the behavioral level that more than one viable continuation was available for selection. These invariant couplings are therefore distinguished from informational behavior change, which requires context-dependent switching among viable alternatives relevant to maintaining Energy, Safety, or Reproduction.

Three units of organismal representation
Before turning to the data, one additional clarification is required: when a behavior change occurs, what is the organism adjusting? Not every survival-relevant adjustment is directed at the body itself. In natural systems, organisms routinely regulate extensions and processes that are as critical to persistence as posture or movement. Restricting observation to bodily locomotion alone risks overlooking a substantial portion of information-sensitive adjustment.
For this reason, we distinguish three units of organismal representation (i.e., externally observable targets of adjustment), any of which may serve as the primary target of a behavior change:
Main unit — the organism itself.
This includes the most frequently observed switches: postural shifts, reorientation, approach/avoidance, strike/withdrawal, reveal/conceal. A gecko’s freeze → flee, a shark’s patrol → pursuit, or a bird’s display → concealment are all examples of Main-unit behavior change.
Vital unit — survival infrastructure.
This includes nests, burrows, dens, webs, dams, lodges, lairs, territories, and defended spaces. Even in the absence of constructed structures, spatial control can function as an actionable asset: patrol, defend, expand, abandon, or relocate. Adjustments at this level often determine long-term safety or energetic stability rather than immediate bodily position.
Reproductive unit — reproductive partners and substrates.
This includes mating partners, gametes, eggs or offspring, and structures or locations used specifically for reproduction. Behavior changes such as courtship displays, pair-bonding maneuvers, clutch guarding, brooding, provisioning, or coordinated care belong to this unit.
These units are not mutually exclusive. A single episode may involve multiple targets (e.g., Main + Reproductive when a parent shields offspring; Main + Vital when defending a burrow). Units may also shift with context: a nest functions as a Vital unit outside the breeding season and as a Reproductive unit during incubation.
This distinction is particularly important for organisms whose bodily movements are sometimes described as mechanically constrained or stereotyped. Even in such cases, behavior change may be expressed through the initiation, delay, suppression, or redirection of reproductive or infrastructural activity, rather than through overt locomotion. By tracking which unit is being regulated, behavior change can be identified without assuming internal mechanisms or invoking debates about reflexes, tropisms, or cognition.
For the purposes of this study, any behavior change that targets one or more of these units, and that plausibly serves to maintain Energy, Safety, or Reproduction under changing conditions, is treated as a legitimate instance of information-sensitive adjustment.

The Dataset: scope and structure
The dataset [DOI: https://doi.org/10.17605/OSF.IO/VB2NC] summarizes 1,530 species across major clades. Each row captures a lineage’s behavior-change profile (B1→B2), mapped to five domains of informational relevance (Tasks 1–5), with partial vs. full maturity, conservative group placement, and basic ecological/social context. The schema is designed to be machine-readable and human-readable.
Disclaimer. The dataset is an evolving document; some figures and labels may be refined in later releases.

Block 1. Technical codes and naming columns

• Species_Code — internal unique ID for the species entry (runs from sp0001 to sp1530).
• Item_Code — unique ID composed of the batch and the species (e.g. b087_sp0867); used when the same species appears in more than one batch/context.
• Batch_Number — numeric batch identifier (001–153); a batch is a curated set of species representing a relatively homogeneous family, clade, or life-strategy group.

Block 2. Species labels

• Unified_Batch_Label — human-readable batch tag combining clade/lifestyle (e.g. “Mammalia — Echolocators (reciprocity & coordinated foraging)”).
• Species_Label — canonical species name (e.g. Panthera leo).
• Clade_Label — higher grouping string (e.g. “Chordata — Mammalia (Carnivora: Felidae)”).
• Taxon_Label — lowest reliable taxon when the species is uncertain (genus/family level).
• Common_Label — common name, if available (e.g. “lion”).

Block 3. Domains and tasks

• BCBS_1 … BCBS_5 — stand for behavior-change / controller maturity for Tasks 1–5. The code is [T][MMM], where: T = task/domain index (1–5), MMM = maturity level: 000 = no credible evidence; 033 = rudimentary / proto control; 050 = moderate control; 066 = strong but not universal control; 100 = routine, species-typical (full) control. Examples: 1100 = Task 1, full; 2033 = Task 2, rudimentary; 3050 = Task 3, moderate; 4066 = Task 4, strong; 5100 = Task 5, full, 5000 = Task 5, no credible evidence.
• task_full — highest task routinely solved (1–5); mirrors the highest BCBS that equals 100.
• task_partial — highest task with credible partial maturity (1–5) — i.e. strongest non-100 value (033/050/066).
• task_domain — primary domain of informational relevance derived from task_full / task_partial
• (1 = Environment & fragments; 2 = Free-moving entities at a distance; 3 = Perception-shaping & signalling; 4 = Group dynamics & coalitions; 5 = Formal symbolic systems).

Block 4. Groups and stacking

• group_min — conservative floor group assignment; equals task_full (for highest task_full =100).
• group_max — conservative ceiling that allows the recorded partial; equals task_partial (for any task_partial>0).
• group_span — shows whether the lineage is transitional (e.g. 2–3, 3–4, 4–5) or non-transitional (1–1, 2–2, … 5–5).
• Group_Name — readable label (e.g. “Group-1 (Binary)”, “Group-2 (Elementary)”, “Group-3 (Manipulatory)”, “Group-4 (Combinatory)”, “Group-5 (Symbolic–Sapient)”; transitionals named accordingly, e.g. “Group 2→3 (Elementary → Manipulatory)”). Group_Name labels are mnemonic, not ontological.

Block 5. Notes and context

• Note — free-text qualifiers (e.g. “Task-3 primarily lab-elicited; wild reports rare”; “territorial defense treated as Vital unit”).
• sociality_primary — main social pattern (e.g., solitary, pair-bonded, family-living, group-living, colonial).
• sociality_tags — additional descriptors (e.g., fission–fusion, cooperative breeding, dominance hierarchy, facultative grouping).
• temporal_status — extant / extinct / domesticated / urban-dwelling / invasive / seasonal, etc. 

The dataset structure at a glance 
Read the dataset left to right and it tells a single story. The ID and label columns (Species_Code, Species_Label, Clade_Label) tell you who we are looking at. The BCBS_1–BCBS_5 columns tell you which of the five informational tasks this lineage can actually control, and to what maturity. The task_full / task_partial pair tells you how far up the sequence the species is stably operating, and what is emerging. The group_min / group_max / group_span trio then turns that into a placement on the five-task scaffold (pure Group-2 vs. Transitional 2→3, etc.). Finally, the Note, sociality_primary/tags, and temporal_status fields tell you the ecological and social setting in which those behavior changes are normally expressed. In other words: who it is → what tasks it controls → how far it has climbed in the five-task sequence → in what context that control shows up.
Throughout this paper, “control over Task K” is used interchangeably with “capacity to tackle Task K.” By control we mean a routine, species-typical ability to deploy the relevant behavior change when that domain is live; we do not mean that the organism is always performing that task.

Worked example: Salamander — row walkthrough
Species_Code: sp0867
Item_Code: b087_sp0867
Batch_Number: 87
Unified_Batch_Label: Amphibians — Urodeles — Courtship & Chemosignal Dances
Species_Label: Salamandra salamandra (fire salamander)
Clade_Label: Chordata — Amphibia (Urodela: Salamandridae)
Taxon_Label: Salamandra salamandra
Common_Label: salamander
BC maturity (BCBS_1–5):
BCBS_1 = 1100 → full control at Task 1 (environmental states & fragments)
BCBS_2 = 2100 → full control at Task 2 (free-moving entities at a distance)
BCBS_3 = 3066 → strong but not universal control at Task 3 (perception-shaping & signalling)
BCBS_4 = 4000 → no credible control at Task 4 (group dynamics & coalitions)
BCBS_5 = 5000 → no credible control at Task 5 (formal symbolic systems)
Derived grouping fields:
task_full = 2 (highest task with 100-maturity)
task_partial = 3 (highest task with partial maturity)
group_min = 2
group_max = 3
group_span = 2–3
task_domain = 2–3 (Task 2: free-moving entities at a distance; Task 3: perception-shaping & signalling)
Group_Name = Elementary → Manipulatory (transitional)
Notes on what this row tells us:
This species is fully at Task 2, is already doing Task-3-like courtship/signal work strongly (3066), but does not yet show group-level coordination (Task 4). So it sits exactly where we expect many amphibians to sit: above pure Task-2 hunters with strong but not full manipulatory capacity at Tasks 3, and below true Task-4 — social coordinators.

Coding protocol and assignment logic
Species-level task assignments were derived inductively from documented behavior-change patterns rather than from a priori task labels. The basic observational unit was a controlled behavior change (B1→B2) triggered by a non-contact environmental event or configuration and plausibly involving a selection among alternatives relevant to Energy, Safety, or Reproduction (ESR). For each species, we surveyed high-consensus natural history sources for recurrent, species-typical instances of such behavior change across ecological contexts. Coding proceeded in three steps. 
First, documented behavior-change episodes were identified and tagged by their primary domain of informational relevance (e.g., environmental state, free-moving entity, audience perception, group structure, or symbolic convention), independent of species identity or phylogenetic position. 
Second, for each domain, we assessed whether the corresponding task was expressed as a routine, load-bearing capacity of the lineage or only in restricted, context-dependent, or atypical forms. This distinction was captured using graded maturity codes (000–100), reflecting absence, partial emergence, or species-typical control. These graded maturity codes are intended as a preliminary, coarse-grained assessment of non-full task expression, used primarily to identify transitional lineages and to avoid forcing categorical placement where capacities are context-limited or unevenly expressed. They do not constitute a fine-grained ranking of species, nor do they affect the ordering, sequencing, or definition of the five tasks themselves. Further refinement of partial-task expression would require task-specific operationalization beyond the scope of the present survey. 
Third, species were grouped conservatively according to the highest task they reliably controlled, with adjacent transitional spans (e.g., 2→3, 3→4) used to represent partial or emerging capacities rather than forcing categorical placement. Importantly, task definitions were not used to pre-sort species. Instead, task categories emerged as minimal, recurring abstractions that summarized which kinds of informational situations reliably elicited behavior change across lineages. 
This procedure is analogous to survey-based segmentation in the social sciences, where latent dimensions are inferred from response patterns rather than imposed in advance. Ambiguous or weakly supported cases were coded conservatively as partial and annotated, and behaviors reported only from highly artificial or single-individual contexts were not treated as full task control.

Data provenance
All species profiles in this dataset were assembled from high-consensus natural history sources: comparative ethology handbooks, field and museum-style species accounts, curated encyclopedia entries, conservation/behavioral assessments, and other widely cited secondary syntheses. Where multiple sources converged on the same description of a behavior-change pattern, we treated that pattern as species-typical. 
Coding was intentionally restricted to behavior that is repeatedly described as routine in a lineage’s normal ecology (courtship sequences, territorial signalling, coordinated mobbing, cooperative breeding, alarm recruitment, clutch guarding, coalition support). Behaviors reported only from narrow laboratory protocols, single exceptional individuals, or highly artificial settings were not treated as full capacities for that lineage; those were coded as partial and/or annotated in the Note column. 
For key theoretical scaffolding and classic descriptions of context-sensitive action, see Selected Works Consulted.

Results

Tasks, domains, and comparability across species
Each informational task identified in this study is defined by the type of situation that reliably precedes ESR-relevant behavior change, rather than by specific behaviors, anatomical structures, or presumed internal mechanisms. Tasks are therefore indexed to domains of informational flow—environmental states, free-moving entities, perception shaping, group structure, and symbolic systems—rather than to taxon-specific repertoires.
This operationalization allows lineages with radically different morphologies, ecologies, and biological substrates to be compared on a common footing: by the kinds of situations that force them to change what they are doing. In the sections that follow, we summarize each task, its corresponding domain of informational control, and the characteristic behavior-change patterns through which that control becomes observable.

Group 1 — Task 1: Binary (0–1) orientation — Environmental control

Domain of informational relevance: Environmental states and fragments.
The immediate physical setting and its fragments: gradients, shelter/exposure, substrate changes, microclimate, barriers, vibration fields, current/flow direction, etc.
Shared capacity (potential)
Species in Group 1 can detect and control changes in this domain. Without being struck or forced, they register changes in the immediate setting that reliably precede behavior change. In this sense, features such as light, temperature, substrate, cover, and moisture function as informational cues.
Foundational decision layer — options to be chosen
Choosing among simple, binary alternatives: favourable / not favourable; safe-enough / not safe-enough; stay / leave; process / pause.
Behavior change patterns (characteristic switches)
These lineages reliably perform controlled shifts such as: move → stay, hold → relocate, drift → anchor, activate → deactivate. Each is a spatial or exposure alternative (here vs. there; now vs. not-now) deployed in service of energy, safety, or reproduction. Accordingly, at this level, informational control is identified solely through observable selection among viable behavioral continuations—expressed through regulated changes in timing, rate, direction, or engagement—rather than through direct physical forcing or invariant coupling, in accordance with the general inclusion and exclusion criteria defined above.
Clarifying the boundary between Task 1 and non-task processes
Below Task 1 lie physical and chemical processes that change state only through direct interaction or force, without the capacity to select among alternatives. Such processes do not produce behavior change in response to environmental change, and therefore cannot act to align conditions with Energy, Safety, or Reproduction. Task 1 marks the minimal transition at which a system can regulate its own behavior in response to changing surroundings in order to sustain itself over time.
Decision cue — what this task is really asking
Informational question the organism is tackling:
“Should I stay or should I move?”, “Is this state safe enough to continue?”, “Do I activate or suspend what I’m doing?”
What this task does not require
It does not require tracking another moving organism, signalling to an audience, coordinating with group members, or following a learned social rule. This is not coalition behavior, display strategy, or symbolic interpretation. It is direct regulation of local conditions.
Representative lineages
All extant taxa in the dataset express this capacity at full strength—from prokaryotes to vertebrates. Every living species is coded BCBS_1 = 1100 for Task 1. At the same time, there is a subset of lineages whose survival appears to rely on this task and essentially this task alone—bacteria, many unicellular eukaryotes, plants, fungi, and comparable forms.

Group 2 — Task 2: Distal engagement choice — Free-moving-entity control
Domain of informational relevance: Free-moving entities at a distance
Motile organisms in the surroundings that can approach, flee, pursue, threaten, court, or otherwise act independently (prey, predators, rivals, mates, intruders).
Shared capacity (potential)
Species in Group 2 can control behavior in response to these free-moving agents at range. They do more than notice “the environment changed” (Task 1). They track a particular moving other, treat it as meaningful “for me, now,” and select among structured options: close in, shadow, confront, evade, hide, or ignore. In other words, they can make decisions about that moving entity before direct contact.
Foundational decision layer — options to be chosen
Distal agent-level alternatives: prey vs. predator vs. rival vs. mate vs. irrelevant intruder. 
Once an entity is differentiated in this way, a corresponding class of behavior change becomes available.
Behavior change patterns (characteristic switches).
These lineages reliably show dyadic, target-directed switches such as: forage → flee, patrol → pursue, approach → veer off, stalk → abandon, intercept → break contact, chase → avoid, trail → dodge. Each is a B1→B2 that is about a specific moving other, not about the backdrop.
Decision cue (what this task is really asking)
Informational question the organism is tackling: “What do I do about that moving entity at a distance?” This is already qualitatively different from Task 1. The organism is no longer only regulating itself against a setting (“Is this place okay?”). It is regulating itself in relation to an independently moving agent at range; it is “I must do X about that mover.”
Clarifying the boundary between Task 2 and Task 1
In Task 1, behavior change is selected to regulate the organism’s relation to its immediate setting (stay/leave, activate/suspend) without reference to any specific moving other. In Task 2, behavior change is selected in relation to a particular free-moving entity, requiring the organism to track, engage, avoid, or disengage from that entity at a distance. Put simply, Task 1 governs whether and how I adjust to the situation; Task 2 governs what I do about that mover.
What this task does not require

• Perception-shaping or audience design (that is Task 3).
• Coalition, role, or recruitment logic (that is Task 4).
• Formal rule-following, symbolic media, or cumulative culture (that is Task 5).

Clear, stable Group-2 exemplars in the dataset include

• Many solitary and reef-associated fishes (including many sharks and predatory teleosts) that patrol, detect, trail, lunge, or disengage purely on the trajectory/speed/proximity of another swimmer, but do not routinely use adaptive camouflage, display choreography, or signalling to shape the other’s perception.
• Turtles and many salamanders/newts that react to detected movers (approach or withdraw) but do not regularly modulate what the other sees.
• Non-social, visually hunting insects, non-displaying lineages such as many dragonflies/damselflies and tiger beetles, and host-seeking dipterans such as mosquitoes and horse flies, and other species that track, close in, or break off based on the motion of a target, but do not, as a lineage-level routine, add perception-shaping displays.

Representative lineages
In coding terms, Group-2 species are those with:

• BCBS_1 = 1100 (full Task-1: Binary environmental control), and
• BCBS_2 = 2100 (full Task-2: free-moving-entity control).

Transitional 2→3 cases (e.g. strong but not universal signalling/decoy/display in some contexts) are coded separately and do not alter the five-group scaffold.

Group 3 — Task 3: Perception-shaping in context — Perception-shaping & signalling control
Domain of informational relevance: Context, contextual cues, signals, and audience perception
The domain here is not “the world out there” (Task 1) and not “a mover out there” (Task 2), but the perceptual channel between organism and observer. The organism adjusts what others perceive—via display, concealment, mimicry, exaggeration, posture, coloration, movement pattern, timing, or signal emission—in order to alter the other’s interpretation before contact.
Shared capacity (potential)
Species in Group 3 can control how they appear to others, actively manipulate others’ perception of themselves or the situation, and, when needed, steer observers toward a particular interpretation or direction of action. They do more than regulate themselves against the backdrop (Task 1) or decide how to engage a moving other (Task 2). They can modulate how they are detected through the visual, auditory, or chemosensory channels of another organism. That includes attraction (“notice me as a mate / as dangerous / as cooperative”), deception (“do not notice me,” “misread me,” “treat me as something I am not”), and direction/steering (“move where I point you”). The key is that the organism’s behavior change is directed at altering what is perceptually available to others.
Foundational decision layer — options to be chosen
Given a situated audience (predator, prey, mate, rival, intruder), select among: attract vs. camouflage; show vs. hide; exaggerate vs. downplay; signal vs. stay silent; mimic vs. reveal true form; lure closer vs. push away; direct this way vs. that way.
Behavior-change patterns (characteristic switches)
These lineages reliably perform BCs of the form: neutral posture → threat display; cryptic stillness → courtship dance / color flare; foraging posture → warning exaggeration (spines out, body inflation, wing spread); background-match camouflage → sudden contrast flash (startle, misdirection); ordinary movement → luring movement (prey-attraction feint, brood-defense decoy); freeze → motion-jerk to trigger pursuit; ordinary presence → directional cueing (gestures/movements to steer conspecifics).
Decision cue (what this task is really asking)
Informational question the organism is tackling: “What do I make you think I am right now?” rather than only “What do I do about you?” It also covers questions of influence, such as: “How can I act so that your behavior changes the way I need?” and “How can I present myself so you interpret this situation in the direction I want?” That is the diagnostic leap from Task 2 to Task 3. In Task 2, the organism chooses how to deal with an agent. In Task 3, the organism chooses how the agent will read the situation. Importantly, the same individual can alternate hide → advertise → hide again, depending on who is present and what is at stake (mate nearby vs. predator nearby vs. offspring nearby). That context-sensitivity of presentation is the signature of Task-3 control.
Clarifying the boundary between Task 3 and Task 2
In Task 2, behavior change is selected to regulate how the organism engages a free-moving entity (approach, avoid, pursue, disengage). In Task 3, behavior change is selected to regulate how that entity interprets the situation, by selectively shaping what it perceives about the organism or the context. Put simply, Task 2 governs my action relative to you — “What I do about you”; Task 3 governs your interpretation because of me — “What I make you think is happening”.
What this task does not require

• It does not require coalition or shared role structure (Task 4). All of this can be done dyadically or one-to-many.
• It does not require symbolic rule systems, language, money/tokens, or maps (Task 5).
• It does not count always-on morphology as Task 3. Aposematic coloration that is permanently visible is a carrier, not a choice. Task-3 evidence requires selectable presentation: show / withhold / transform based on context. This avoids the slide from “feature visible” to “feature chosen.”

Representative lineages

• Many reptiles and amphibians that switch between crypsis and display (cryptic posture → sudden throat fan, color flush, or body inflation to deter or attract).
• Numerous birds and fishes that perform courtship displays, alarm postures, or fin/spine exaggerations on demand, not continuously.
• Insects that deploy deimatic/startle flashes, wing-spot reveals, or mate-attraction signalling only in specific contexts—not as an always-on message.

Coding interpretation
Group-3 species express:

• BCBS_1 = 1100 (Task-1: Binary orientation — Environmental control)
• BCBS_2 = 2100 (Task-2: Distal engagement choice — Free-moving-entity control)
• BCBS_3 = 3100 (Task-3: Perception-shaping & signalling control)

Lineages with BCBS_3 = 3066 (strong but not universal Task-3) are coded as Transitional 2→3 or 2–3 span in Appendix A. These transitional cases are important: they show that Task-3 style perception-shaping emerges in partial, context-bound form before it becomes a universal, load-bearing strategy for the lineage.

Group 4 — Task 4: Coalition alignment — Group-dynamics control
Domain of informational relevance: A structured group of independent organisms
The domain here is not a single mover (Task 2) and not just an audience to manage (Task 3), but a multi-individual system in which members are interdependent: they recruit help, share risk, delegate responsibility, allocate roles, maintain or shift hierarchy, defend together, raise young together, enforce tolerance or status, or suppress rivals.
Shared capacity (potential)
Species in Group 4 can control behavior in relation to changes in group dynamics—coalitions, alliances, rank, joint projects. This includes both collaborative and competitive moves inside the group: coordinating movement or attack; supporting an ally and not a rival; tolerating one juvenile and rejecting another; exchanging roles and sharing responsibilities; defending a nest or den as a joint project; recruiting others for mobbing or hunting; deferring to rank when it is strategically necessary. In other words, individuals act in ways that make sense only relative to a live social structure. 
Task 4 groups a diverse set of behaviors not because they are superficially similar, but because they share the same underlying control problem. In all Task-4 cases, behavior change is selected with respect to a structured, multi-individual system in which outcomes depend on roles, relationships, and coalition state rather than on a single other agent or audience. Actions such as alliance support, cooperative defense, parental provisioning beyond the dyad, rank-dependent tolerance or aggression, and recruitment during mobbing differ in form, but converge in that their appropriateness depends on who is involved, how individuals are related, and how the group is currently organized. This distinguishes Task 4 from Task 3, where behavior is conditioned on shaping how others perceive the situation, and from Task 5, where behavior is regulated by abstract, symbolic conventions. The apparent heterogeneity of Task-4 behaviors therefore reflects the breadth of group-structured contexts rather than a lack of specificity in the task itself.
Foundational decision layer — options to be chosen
Load-bearing choices for group maintenance, parental success, offspring survival, or individual access to resources/mating, such as: collaborate vs. compete; recruit vs. do not recruit; help vs. defeat; protect vs. harass; share vs. take; follow vs. lead; reinforce hierarchy vs. challenge it; defend this offspring/resource vs. yield; intervene for A against B vs. stay neutral; admit this individual vs. exclude that one; stay with the group vs. split off.
Behavior-change patterns (characteristic switches)
These switches are routinely observed as:

• forage solo → join coordinated hunt/defense
• ignore conspecific → intervene on its behalf (alliance support)
• tolerate proximity → chase/expel after rank breach
• rest quietly → alarm-call and recruit a mob against a predator or rival group
• tend own young → provision or guard non-offspring in the same unit
• withdraw from fight → re-enter when an ally is threatened
• These are not just “approach/avoid a mover” (Task 2). They are role-aware engagements: you do X because that one is your ally, your cub, a tolerated subordinate, or a coalition partner against an outgroup.

Clarifying the boundary between Task 4 and Task 3
In Task 3, behavior change is selected to influence the perception of an observer or audience, typically in dyadic or one-to-many contexts. In Task 4, behavior change is selected relative to a structured group, where actions depend on roles, alliances, coordination, and competition within a multi-individual system. Put simply, Task 3 governs how I appear to others — “What do I make you think I am?”; Task 4 governs how I act within a coalition — “What do I do given who we are to each other?” Task 3 manipulates perception; Task 4 navigates relational structure.
Decision cue (what this task is really asking)
Informational question the organism is tackling: “Who is us and who is them right now?”, “What position do I take inside this live social structure?”, “What action keeps this arrangement working for me (and mine) in this moment?”, and “What is my role in keeping this structure working for me and mine?” That is qualitatively different from Task 3 (“what do I make you think I am right now?”). Task 3 is performance for an audience. Task 4 is navigation of a network.
What this task does not require

• It does not require formal, externalized symbolic systems (Task 5). Coalition alignment can run without language, writing, maps, money, or law codes.
• It does not require explicit rule-teaching; much of it is enacted through practice, reinforcement, and expectation.
• It does go beyond simple proximity or shoaling. Mere aggregation (schooling fish, loose herds with no role structure) does not count as Task 4. Task 4 requires patterned role-taking, coalition logic, or coordinated joint action where individual fates are entangled.

This distinction matters: being near others is not the same as structuring a coalition with them. For that reason we reserve full Task-4 for systems in which we can see collaboration and competition operating together, and in which role taking, coalition choice, or support decisions are themselves behaviorally selectable by the individual (join/withhold, back A against B, tolerate X but not Y), rather than fixed by developmental caste or chemistry. Highly coordinated insect or fish systems are therefore coded 3→4 or 4-partial, when coordination is strong but coalition recombination or role reassignment is limited at the individual level.
Representative lineages

• Mammals very often express Task-4 style control. Because mammalian life histories are scaffolded by juvenile group interaction, provisioning, protection, and rank/tolerance management, removal of this capacity would severely constrain mammalian life histories as we know them.
• Many cooperative or family-living birds (some corvids, parrots, cooperative breeders) that jointly defend territories, provision young beyond the breeding pair, and recruit group members in alarm/mobbing contexts.
• Some fishes and insects with role differentiation, coordinated defense, or shared nest/colony maintenance. In the dataset, these often sit in 3→4 or 4-partial slots when coordination is strong but internal competition, negotiated alliances, or true coalition choice is still limited. Full Task-4 is reserved for cases where coalition alignment and differentiated roles are both central and behaviorally selectable.

Coding interpretation
Stable Group-4 species express:

• BCBS_1 = 1100 (Task-1: Binary orientation — Environmental control)
• BCBS_2 = 2100 (Task-2: Distal engagement choice — Free-moving-entity control)
• BCBS_3 = 3100 (Task-3: Perception-shaping & signalling control)
• BCBS_4 = 4100 (Task-4: Coalition alignment — Group-dynamics control)

Species that show coalition-like organization but not universally (e.g. BCBS_4 = 4066) appear as Transitional 3→4 in Appendix A. These transitional spans (3–4, 4-partial) are exactly where we see cooperative hunting, communal defense, extended parental care, or group-structured rank negotiation present, but not yet obligatory for all members of the lineage.

Group 5 — Task 5: Rule-guided abstraction — Formal symbolic systems control
Domain of informational relevance: Formalized symbolic systems and abstract conventions (“Symbolic–Sapient”)
Population-level systems of shared symbols—language, number, writing, money/tokens, mapped space, explicit norms, ritual contracts, durable collective memory—that can prescribe action and coordinate expectations even when the relevant individuals or objects are not directly present.
Shared capacity (potential)
Species in Group 5—in the present dataset, at full maturity humans only—can control behavior using abstract, conventional symbols and structured rule systems. Behavior can now be selected and switched not only because “the environment changed” (Task 1), or “that agent is doing something I must address” (Task 2), or “I want to shape how you see me” (Task 3), or “our coalition is in state X so I take role Y” (Task 4), but because “a shared symbolic rule says we now do Z,” or “the map/plan/story says go here,” or “this token changes ownership/obligation,” or “this declaration updates the social state.” In other words, behavior change can be driven by symbols that stand for absentees, elsewheres, and non-immediate states of affairs.
Foundational decision layer — options to be chosen
Given a shared symbolic frame, the organism can decide to:

• follow an instruction, policy, law, promise, or norm that is only symbolically present;
• act on a plan encoded in language or marks (“meet at sunrise at that rock”);
• accept or transfer obligations via a token (currency, promissory object, seal, record);
• update group state by declaring something in language (“we are at peace now,” “you are in/out”);
• generate novel combinations (new sentences, new marks, new routes) and have them understood as binding by others. These are not just “make ally / don’t make ally.” They are “instantiate a convention, then behave because of that convention.”

Behavior change patterns (characteristic switches)
Routinely observed human-style B1→B2 include:

• idle → comply based on an instruction, rule, or agreement (no immediate physical force);
• local foraging → coordinated task-execution based on a shared, verbally stated plan;
• hold resource → transfer resource because of an abstract token (“this paper now means it’s yours”);
• short-horizon action → long-horizon investment because of a symbolic narrative, model, schedule, or map;
• spontaneous motion → ritualized behavior because “this ceremony/norm now says we do X.”
• In all such cases, a symbolic frame can trigger or suppress action even when no immediate cue, rival, mate, ally, or predator is present.

Clarifying the boundary between Task 5 and Task 4
In Task 4, behavior change is coordinated through live group dynamics, roles, and relationships among present individuals. In Task 5, behavior change is coordinated through abstract, shared symbolic systems that can prescribe action independent of immediate presence, perception, or interaction. Put simply, Task 4 governs alignment through social structure; Task 5 governs alignment through formalized rules and symbols.
Decision cue (what this task is really asking)
Informational question the organism is tackling: “How do I act now because of a shared symbol/rule that is not physically here?”, “How do I use our symbolic system to align, coordinate, and bind others (and myself), “How do I use our symbolic system to maintain coordinated expectations across time and space?”
This is qualitatively beyond coalition. In Task 4, alignment depends on who is present and what the coalition is currently doing. In Task 5, alignment can depend on a rule that outlives the situation and can be invoked later by anyone who knows the system.
What this task does not reduce to

• Not just rich signalling or ritual display. Many birds, cephalopods, reptiles, and mammals have elaborate, context-sensitive signalling—that is Task 3 (and, in social settings, Task 4). Task 5 requires conventions that can be followed, referenced, revised, and transmitted as systems, not just emitted moment by moment.
• Not just social coordination. Many mammals and birds coordinate hunts, raise young cooperatively, or enforce rank—that is Task 4 (Group-dynamics control). Task 5 requires explicit, shared symbols that can organize behavior beyond immediate presence or immediate memory.
• Not “an animal in a lab can press a symbol.” We explicitly distinguish an individual’s lab-trained potential from a population-level practice. Full Task-5 requires that the symbolic system is normal, entrenched, and socially scaffolded for that lineage outside the lab.

This last boundary matters. Comparative cognition has clearly shown that great apes, corvids, parrots, cetaceans, and elephants can acquire symbols, transmit practices, and use context-sensitive signals. What we do not yet see outside humans is a population-level, obligatorily shared, and consciously revisable symbolic system that organizes day-to-day coordination across time and space. For that reason we code these lineages as Task-5 partials (5033, 5050, 5066) and place them in transitional 4→5 slots, rather than in the full Group-5 class.
Representative lineages
Humans are the only lineage assigned full Task-5 control (BCBS_5 = 5100), because routine human life depends on language, shared fictions, money/tokens, mapped abstractions, explicit norms and laws, and consciously revisable social contracts.
Some other vertebrates (great apes, corvids, parrots, cetaceans, elephants) exhibit partial Task-5 signatures in restricted contexts—symbol learning, proto-conventional calls, tool traditions with social transmission. In the dataset these are coded with partial maturities (e.g. 5033, 5050, 5066) and placed in Transitional 4→5, not in full Group-5.
Task 5 is defined by control over formal symbolic systems that regulate behavior through shared conventions rather than immediate context, presence, or direct interaction. This includes population-level use of symbols that persist across time and space, prescribe obligations, and coordinate action among individuals who may not be co-present. On this definition, humans are the only lineage in the present dataset that exhibits full, routine control of Task 5 as a load-bearing feature of daily life. Importantly, this does not imply an absence of Task-5-related capacities in other species. Several non-human lineages show partial or context-limited signatures of symbolic mediation, such as socially transmitted conventions, learned symbolic associations, or proto-normative signals. These cases are explicitly coded as partial Task-5 expression rather than excluded. The distinction between partial and full control is therefore empirical rather than definitional: full Task-5 control is assigned only where symbolic systems function as a primary, population-wide regulator of behavior, not merely as an individual or experimental capability.

Summary of task structure and constraints
Across the full dataset of 1,530 species, task control patterns exhibit two consistent empirical constraints. First, we did not observe stable cases in which a lineage routinely expresses control over a later task while lacking reliable control over any earlier task in the sequence. For example, no species shows regular Task-3 (perception-shaping) control without also expressing Task-2 (distal engagement) control, and no species shows Task-4 (coalition alignment) control without Tasks 1–3 in place. Second, among lineages expressing multiple tasks, the internal ordering of tasks is conserved: when several tasks are present, they appear in the same ordinal sequence across taxa. We did not observe alternative or scrambled combinations such as [Task-1, Task-3, Task-4] or [Task-2, Task-4] in routine, species-typical behavior. Transitional cases occupy only adjacent spans [e.g., 2→3, 3→4, 4→5], indicating partial or context-limited emergence of the next task rather than bypassing or reordering. These constraints are descriptive properties of the dataset and hold across major clades, ecological niches, and social systems represented here.
Building on these observed constraints, the five domains of informational control sort lineages into five major groups.

On the number of tasks
The Five Task Model is not proposed as an a priori partition, but as a compact descriptive scaffold that emerged from the dataset. Across 1,530 species, the informational situations that reliably preceded behavior change clustered into five recurrent domains. Fewer domains proved insufficient: collapsing these categories obscured stable differences between distal engagement, perception-shaping, and group-level coordination. Conversely, attempts to introduce additional task domains did not reveal new, non-redundant classes of behavior-change control that were both species-typical and irreducible to the five identified here. In this sense, five represents a minimal sufficient set for the present material, not a claim of theoretical finality. 
The framework is explicitly open to falsification: the identification of a stable, species-typical behavior-change domain that cannot be decomposed into or composed from these five would require revision of the model. Questions about why this structure appears to stabilize at five, and under what evolutionary conditions each task became load-bearing, are addressed in Part II.

Conclusion and Discussion 
In this work we asked a very external question: what kinds of informational situations reliably make an organism change what it is doing?
We treated behavior change as that observable switch from one ongoing pattern (B1) to another (B2) that is not caused by direct physical force and that looks like a choice among at least two viable alternatives (stay vs. leave; pursue vs. hide; display vs. conceal; recruit vs. abstain). That switch was our practical footprint of information use.
Each such switch can be read as an attempt to keep Point A — what just changed in the surroundings — aligned with Point B — what must be preserved for survival: energy, safety, and reproduction (ESR). In the wild, that alignment is enforced through timely adjustments in location and interaction; there is no other tool. Seen this way, survival is not just “having behaviors,” but controlling when to change behavior so that A and B remain in register. If that alignment repeatedly fails, the lineage fails. If it keeps holding, the lineage persists.
We then coded these switches across 1,530 species representing major families and lineages across life on Earth. What we found was not an open-ended chaos of triggers. The informational events and situations that precede ESR-relevant switches clustered into a small, repeated set. We described that set as five informational domains, with five corresponding adaptive tasks:

1. Binary (0–1) orientation — Environmental control
2. Distal engagement choice — Free-moving-entity control
3. Perception-shaping in context — Perception-shaping & signalling control
4. Coalition alignment — Group-dynamics control
5. Rule-guided abstraction — Formal symbolic systems control

We are not presenting these five as five instincts or five sealed modules. We present them as five guide rails that appear to organize how very different lineages handle the informational problems that matter most. In the dataset we called this scaffold the Five Task Model. It behaves like a hidden infrastructure of species’ survival and reproduction: it quietly partitions species into five groups, depending on how many of these tasks they can reliably control in daily life.
Three properties of the observed task structure are especially striking.
Stacking across groups.
Species do not scatter randomly across the task space. Instead, they fall into five groups that differ by how many informational tasks they can stably control in daily life. Some reliably solve only the most basic task; others solve two, three, or four; and, in full form, only humans routinely solve all five. Transitional cases appear between these groups (e.g., “2→3,” “3→4”), but they do not introduce new tasks and do not disrupt the underlying scaffold.
No skipping (cumulative inclusion).
Across the dataset, we do not observe stable lineages that routinely express a later task while lacking a required earlier one. There are no consistent profiles such as 1–3–4 (skipping Task 2), 1–2–4 (skipping Task 3), or 1–2–3–5 (skipping Task 4). Later tasks appear to build on earlier ones rather than replace them, yielding a cumulative structure.
No reordering (fixed progression).
We also do not observe reshuffled task sequences. Where a later task is in regular use, the earlier tasks appear in place and in the same order. In this sense, “no skipping” means that later tasks do not appear without earlier ones present, while “no reordering” means that the same set of tasks does not appear in a different internal arrangement.
Taken together, these constraints suggest a gated sequence: later control capacities sit on top of earlier ones rather than branching off independently. This implies that the five tasks are doing more than naming surface behaviors. They appear to outline a minimal set of informational control problems that any lineage must solve, in order, to keep environmental change aligned with Energy, Safety, and Reproduction over time. In this sense, the tasks function less like a checklist and more like the load-bearing rails of information use across life.

Seen in this light, differences across living systems can be described not primarily in terms of behavioral richness or biological complexity, but in terms of which informational distinctions they are able—and required—to resolve under constraint. The Five Task structure captures this progression: each additional task corresponds to an expansion in the range of situations that can force behavior change, while preserving earlier constraints. 
Taken together, these observations define what we call the Five Task Model. In practical terms, the Five Task Model represents a species’ information-processing world as a structured set of five elements: 

• five domains of informational flow, 
• five basic informational tasks, 
• five corresponding control potentials, 
• five characteristic behavior-change patterns, and 
• five major groups of species, each group defined by how many of these tasks it can stably solve.

This article (Part I) stayed deliberately outside the organism. We watched when behavior changes, and we mapped which informational pressures most often force that change, across life. The pattern that emerged was unexpectedly compact: the same five-part structure just named—no skips, no re-orderings. In Part II we move one level in. We ask when each task became not just present but load-bearing for whole clades (provenance vs. prevalence), what the biosphere would look like if you “removed” one task at a time, and why this stack seems to stop at five rather than collapsing to four or proliferating to six or nine.

Data Resource
The dataset supporting the findings of this study is openly available via the Open Science Framework (OSF):
Frolov, S. A. (2025). The Five Task Model: From Cognition and Evolution to AGI (Dataset_Species_Domain_Task.csv).
DOI: https://doi.org/10.17605/OSF.IO/VB2NC

Ethical Statement
This study is based exclusively on secondary analysis of published natural history and ethological sources. No new experiments were conducted, and no animals or human participants were involved. Ethical approval was therefore not required.

Published as a Preprint: DOI: https://doi.org/10.17605/OSF.IO/E6BQA

Publication date: January 19, 2026