The Computational Temptation

Computational language and computational methods are deeply embedded within contemporary biology. Genes may be described in programmatic terms, neural activity analysed computationally, regulatory systems modelled through state transitions or dynamical operations, and organisms studied through structured transformations of signals, states, and behaviour.

These approaches have considerable scientific value. Computational models and explanations are used across neuroscience, systems biology, bioinformatics, developmental modelling, artificial intelligence research, and many other areas of biological inquiry.

The term computation, however, does not designate a single homogeneous theoretical framework. Computational approaches can differ substantially in whether they emphasise formal operations, physical implementation, dynamical organisation, representation, distributed processing, algorithmic structure, or other forms of state transformation. Computation should therefore not be identified automatically with digital symbol manipulation, stored programs, central processors, fixed algorithms, or conventional computer architecture.

APS does not reject computational explanation.

Its claim is narrower: APS does not define life as computation or computational organisation.

Within APS:

Life is viability-oriented, constraint-closed organisation.

A living system may contain processes that are usefully described or explained computationally without computation thereby constituting the APS definition of life.

This distinction is one of non-identity. It does not establish that computational explanations are merely descriptive, formal, syntactic, externally interpreted, or biologically incomplete. Nor does it establish that APS provides a deeper explanation simply because it makes viability-oriented organisation explicit.

Where computational and APS explanations address the same biological explanandum, their relative explanatory contribution must be established through target-matched comparison.

Computation and Information Processing

Computation should not be treated as synonymous with information processing.

Information-processing approaches can concern the detection, transformation, coordination, storage, or use of differences within systems. Computational approaches are more specific in the kinds of operations, transformations, architectures, or formal relations they attribute to the processes under investigation.

The exact relation between computation and information processing depends upon the explanatory framework being used. Some computational accounts may also be informational; some informational descriptions need not amount to computational explanations. Neither concept should automatically be identified with representation, cognition, or control.

This distinction matters because biological systems can be analysed under more than one explanatory description. A regulatory process may be described informationally, computationally, dynamically, mechanistically, or organisationally, depending upon the question being asked and the relations being investigated.

APS introduces a different substantive concern. It defines life as viability-oriented, constraint-closed organisation and asks how that organisation maintains and re-establishes continuity through change.

That difference does not make computational or informational explanations secondary by definition. Nor does it show that they merely operate within an organisational reality that APS alone explains.

Where computational, informational, and APS accounts address different explananda, they need not compete. Where they address the same explanandum, their explanatory relationship must be assessed rather than inferred from their terminology.

What Computational Approaches Explain

Computational approaches provide substantive explanatory and modelling resources for many biological phenomena. Depending upon the framework and explanandum, these may include:

  • pattern recognition
  • signal coordination
  • learning
  • decision procedures
  • neural processing
  • regulatory dynamics
  • adaptive behaviour
  • optimisation
  • feedback relations
  • structured transformations of biological states

Computational models can reveal regularities, test hypotheses, specify relations among variables, clarify possible mechanisms, and support explanations of how biological activity changes over time.

Their explanatory contribution should not be reduced to description merely because they use formal representations or abstractions. In particular cases, a computational explanation may be substantive, sufficient, or preferable.

Computational approaches are also heterogeneous. A computational account of neural activity need not have the same architecture as an account of gene regulation, developmental dynamics, bacterial behaviour, or artificial computation. No single model of computation should therefore stand for computational explanation as a whole.

APS approaches some of these phenomena from a different explanatory perspective. It asks how processes participate in viability-oriented, constraint-closed organisation and how such organisation maintains and re-establishes continuity through change.

That is the APS problem of organised persistence.

Organised persistence is not a deeper reality that computational explanation automatically presupposes, and it is not a second definition of life. It is an APS explanatory problem whose contribution must be assessed when it overlaps with the explanandum addressed by a computational account.

Where computational and APS explanations address different questions, no direct competition follows. Where they address the same question, their relationship may be complementary, overlapping, independent, competing, redundant, computation-favouring, APS-favouring, qualified, or null.

Computation and Living Organisation

Computational explanations can be applied to processes occurring within living systems, but this does not determine in advance how computation relates explanatorily to living organisation.

APS defines life as viability-oriented, constraint-closed organisation. Within this account, living organisation is ongoing activity through which a system maintains and re-establishes the conditions of its own persistence.

The corresponding problem of organised persistence concerns how that organisation maintains continuity through change.

Computational approaches may address some of the same processes from different explanatory perspectives. They may characterise transformations of states, relations among variables, regulatory operations, distributed processing, adaptive behaviour, or other forms of biological activity. Depending upon the explanandum, such accounts may be explanatory in their own right.

APS therefore need not claim that computation presupposes an organisational explanation supplied by APS. Nor should computational processes be treated merely as implementations of a more fundamental organisational account.

Instead, the relevant question is relational: when computational and organisational explanations address the same biological phenomenon, what does each explain, what dependencies does each specify, and what explanatory contribution follows from their comparison?

Any proposed dependency between computational processes and living organisation must be specified and assessed rather than inferred from APS architecture alone.

Computation and Biological Normativity

Computational systems and computational explanations can employ criteria governing successful operations, error, optimisation, selection among alternatives, or transitions among states. In engineered systems, such criteria may be externally specified. This familiar case should not, however, be treated as defining computation in general or as determining what computational explanation can contribute in biology.

APS makes a different substantive claim about biological normativity.

Within APS, biological normativity is viability-relative asymmetry: some states, processes, and outcomes contribute differently to the maintenance or loss of the living organisation’s viability.

This normativity is therefore characterised relative to the organisation of the living system itself rather than simply by reference to an externally assigned standard of computational correctness.

The distinction is important, but it does not establish that computational explanations cannot address biologically normative processes. A biological process may be computationally characterised while also participating in viability-relative organisation. Whether the computational account captures, explains, abstracts from, or remains independent of the relevant normative relation depends upon the explanatory target and the particular account being assessed.

APS should therefore not infer from the existence of externally specified computational systems that computation as such lacks resources for explaining biological regulation, adaptive activity, or normatively differentiated processes.

Nor is Biological Evaluation necessarily explanatorily prior to computation. Within APS, Biological Evaluation is the process through which agency generates significance. Where computational processes are proposed to participate in such activity, their relation to evaluation and significance must be investigated rather than fixed by a general dependency ladder.

Biological significance must likewise remain distinct from meaning. Nothing about computational description or explanation, by itself, warrants collapsing computation, information processing, representation, significance, semiosis, cognition, or meaning into a single category.

Computation Without Life

Computation is not sufficient for life as defined within APS.

Artificial and engineered systems provide a straightforward illustration. Systems can perform complex calculations, transform signals, learn statistical regularities, optimise performance, control processes, or generate adaptive outputs without thereby satisfying the APS definition of life.

The relevant conclusion is limited.

It shows that computational capacity alone does not entail viability-oriented, constraint-closed organisation.

It does not show that artificial computation is merely derivative, externally meaningful, or explanatorily superficial. Nor does it establish that computational explanations of biological systems are inadequate.

Engineered computational systems may differ from living systems in their organisation, dependence upon external conditions, modes of maintenance, criteria of operation, and relations to their environments. Those differences must be specified in each comparison rather than attributed to computation as such.

Likewise, the fact that a non-living system can compute does not establish that computation is unimportant to life. A capacity need not define life in order to contribute substantially to explanations of living processes.

Artificial computation therefore supports the article’s non-identity claim:

life is not computation merely because living systems can be characterised computationally.

Nothing stronger follows without further comparative evidence.

Computation Is Not Identical with Program Execution

Some influential computational metaphors portray biological systems in terms of programs, instructions, algorithms, codes, or stored procedures. Such metaphors can be scientifically useful, but they should not be allowed to define computation as a whole.

Living systems undergo continual material turnover, regulation, development, repair, interaction, and reorganisation. Their activity can depend upon histories, changing internal conditions, environmental relations, and dynamically maintained organisation.

For that reason, a model based upon fixed instructions executed by an otherwise stable architecture may be inadequate for some biological explananda.

But this limitation applies to the particular model, not automatically to computational explanation.

Computational approaches need not require a central processor, fixed program, stored symbolic instructions, or an invariant sequence of operations. They may address distributed, dynamical, adaptive, embodied, or physically realised processes.

APS therefore need not decide whether living systems literally “run programs” in order to maintain its central distinction. Its claim is that computational or programmatic characterisation, whatever its explanatory usefulness in a particular case, is not identical with the APS definition of life.

Where a computational model successfully explains adaptive reorganisation or other dynamically changing biological activity, APS cannot reject that explanation merely because the process is computationally characterised.

Constraint Closure and the Organisation of Computation

APS defines life as viability-oriented, constraint-closed organisation.

Constraint closure refers to the organisation through which constraints involved in maintaining the living system are mutually dependent within that organisation. It is part of the APS account of life, not a general criterion by which computational explanations can simply be ranked as incomplete.

Computationally characterised processes may participate in such organisation. They may also provide explanations whose targets do not require an account of constraint closure. The explanatory relation depends upon the biological question being addressed.

APS therefore should not treat constraint closure as automatically grounding, completing, replacing, or providing a deeper explanation of computational organisation.

Nor does abstraction itself count against computational explanation. Scientific explanations routinely abstract from features that are irrelevant to their immediate explanandum. A computational model need not reproduce the entire organisation of a living system in order to provide an adequate explanation of a particular phenomenon.

APS uses Agency, Process, and Scale as complementary analytic projections of one viability-oriented, constraint-closed organisation.

Agency concerns what living systems do.

Process concerns how continuity is maintained despite change.

Scale concerns where persistence is organised across spatial and temporal extents.

These are analytic projections rather than independent components of reality, causal stages, dimensions, levels, or a hierarchy. Their use does not entail that every adequate biological explanation must be multiscale or that computational explanations are deficient when they focus on a restricted spatial or temporal domain.

Their relevance is explanatory and target-dependent.

Within this organisation, biological agency is viability-oriented organisational activity. Computational descriptions or explanations of such activity may illuminate particular operations or relations without thereby becoming definitions of agency or life.

Conversely, APS’s explicit account of agency does not establish that computational explanation is incapable of addressing agentive biological phenomena.

Where APS proposes a dependency between computational activity, agency, constraint closure, or organised persistence, that relation remains a Dependency Hypothesis requiring specification and assessment.

APS and Computational Explanation

APS and computational approaches can address biological phenomena from different, sometimes overlapping, explanatory perspectives.

APS defines life as viability-oriented, constraint-closed organisation. It treats organised persistence as the explanatory problem of how that organisation maintains and re-establishes continuity through change. Within this account, biological agency is viability-oriented organisational activity, and biological normativity is viability-relative asymmetry.

Computational approaches need not adopt these formulations in order to provide substantive biological explanations. They may investigate state transformations, distributed processing, regulatory operations, learning, decision procedures, neural activity, adaptive behaviour, or other biological phenomena according to computationally specified relations.

The existence of these different explanatory resources does not determine their relationship in advance.

A computational explanation may address a target that APS does not directly address. An APS explanation may investigate aspects of living organisation that are outside the target of a particular computational account. Where their targets overlap, the two approaches may identify some of the same dependencies under different descriptions or propose different dependencies requiring comparative assessment.

APS therefore does not incorporate computational explanation as a subordinate component of its own architecture. Nor does computational explanation become biologically intelligible only when situated within APS.

Their relationship may be complementary, overlapping, independent, competing, in tension, redundant, computation-favouring, APS-favouring, qualified, or null.

This openness is methodologically important. APS’s biological specificity, conceptual integration, or explicit treatment of viability, organisation, agency, and persistence cannot by themselves demonstrate explanatory gain over a computational account.

Likewise, the success of computational explanation for a biological phenomenon does not establish that life itself should be identified with computation.

The comparative question is therefore not whether biology is “really” computational or organisational. It is which explanatory relations illuminate the specified biological target, with what evidence, under what conditions, and relative to which alternatives.

Where APS proposes relations among computation, living organisation, agency, viability, or organised persistence, these remain Dependency Hypotheses requiring specification and assessment rather than consequences established by APS architecture alone.

Conclusion

Computational approaches provide important explanatory and modelling resources across biology. Their usefulness is not confined to description, and computation should not be reduced to digital symbol manipulation, conventional program execution, explicit representation, or externally specified control.

APS nevertheless does not identify life with computation.

Life is viability-oriented, constraint-closed organisation.

Organised persistence is the corresponding explanatory problem of how that organisation maintains and re-establishes continuity through change. Neither this definition nor that explanatory problem establishes that computational explanations are subordinate to APS.

The fact that computation can occur in systems that are not alive supports a bounded conclusion: computational capacity alone is not sufficient for life as defined within APS. It does not show that computation is irrelevant to living systems or that computational explanations cannot provide substantive, sufficient, or preferable explanations of particular biological phenomena.

Likewise, differences between computational and APS accounts do not establish which provides the better explanation where they address the same target.

That question requires target-matched comparison.

APS and computational approaches may therefore prove complementary for some questions, independent for others, or genuine competitors where they specify different explanations of the same phenomenon. Comparative assessment must remain capable of yielding computation-favouring, APS-favouring, qualified, redundant, or null results.

The legitimate conclusion of this article is consequently one of non-identity rather than superiority:

life, as defined within APS, is not identical with computation or computational organisation.

Whether APS provides explanatory gain over a computational alternative for any shared biological target remains an empirical and methodological question rather than a consequence of APS terminology or architecture.

Key Point

Life is not identical with computation, but non-identity does not establish the inadequacy of computational explanation. APS defines life as viability-oriented, constraint-closed organisation and treats organised persistence as the problem of continuity through change. Computational approaches may nevertheless provide substantive, sufficient, or preferable explanations of particular biological phenomena. Where computational and APS explanations address the same target, their relative explanatory contribution requires target-matched assessment.

Explanatory Architecture

Central Question

How should computational explanation be related to the organisation of living systems without identifying life with computation or assuming that computational explanations are thereby biologically inadequate?

Architectural Role

This article establishes a non-identity clarification within the APS comparative-theory corpus. It distinguishes the APS definition of life from computational characterisations of biological processes while preserving computational explanation as a potentially substantive and sufficient form of biological explanation.

Its function is not to determine whether APS or a computational approach provides greater explanatory power for any particular biological target.

Preceding Explanatory Dependencies

The article uses the APS definition of life as viability-oriented, constraint-closed organisation and the APS account of biological agency as viability-oriented organisational activity.

It also relies on the distinction between computation and information processing and on the requirement that biological significance remain distinct from meaning.

These are conceptual dependencies within the APS explanatory architecture. They do not establish causal, temporal, ontological, hierarchical, or foundational priority over computational processes or computational explanations.

Any proposed biological dependency between computation, living organisation, agency, viability, significance, constraint closure, or organised persistence remains a Dependency Hypothesis requiring specification and assessment.

Subsequent Explanatory Developments

The distinction established here supports subsequent examination of computation in relation to biological agency, cognition, artificial systems, representation, semiosis, and other forms of biological organisation.

Those developments must not infer cognition merely from computation, information processing, representation, regulation, prediction, or adaptive response.

Nor should they assume that computational processes acquire biological explanatory legitimacy only through incorporation into APS.

Related questions include:

  • When does a computational model provide a substantive biological explanation rather than a useful formal representation?
  • How should computation be distinguished from information processing, representation, control, and cognition in particular biological cases?
  • Can computational explanations capture viability-relative normativity or Biological Evaluation, and when is doing so relevant to the explanandum?
  • How should computational and organisational explanations be compared when they address the same biological phenomenon?
  • Which proposed relations among computation, agency, significance, cognition, and living organisation constitute testable Dependency Hypotheses?

These questions cannot be settled by non-identity alone.

Position Within APS

Within APS, Life is viability-oriented, constraint-closed organisation.

Biological agency is viability-oriented organisational activity.

Biological normativity is viability-relative asymmetry.

Biological Evaluation is the process through which agency generates significance.

Organised persistence concerns how living organisation maintains and re-establishes continuity through change; it is not a second definition of life.

Agency, Process, and Scale are complementary analytic projections of one viability-oriented, constraint-closed organisation rather than independent components, causes, dimensions, levels, or hierarchy.

Computational processes may participate in living organisation and computational explanations may illuminate biological phenomena without computation thereby becoming the APS definition of life. Conversely, APS’s conceptual architecture does not itself demonstrate explanatory superiority over computational approaches.

Where both address the same explanatory target, comparative explanatory gain requires target-matched assessment.