From Object to Structured Representations

Implications of a Contextual Architecture of Scientific Modelling

Science is often framed as the study of objects: particles, systems, agents, variables. We describe what exists, assign properties, and build models that map those properties to outcomes. On this view, understanding the world is a matter of correctly specifying the object.

But in practice, this is not how science actually works.

Across domains, outcomes are not determined by the object alone, but by the object together with the conditions under which it is represented, measured, and inferred. A quantum system behaves differently depending on the measurement basis. A machine learning model produces different outputs depending on the prompt. Economic variables evolve differently depending on expectations and available information. In each case, the same underlying system yields different outcomes under different conditions.

This points to an important shift in how we should think about scientific representation:

from object → to object-as-represented-under-structure

What science evaluates is not, in general, an object in isolation, but an object under a structured set of conditions that determine what counts as admissible, what can be inferred, and what can be observed.

Objects are not enough

It is natural to think that if we specify a system completely, we should be able to determine its behaviour. But many scientific frameworks already show that this is not the case.

The outcome of a measurement, an inference, or a decision depends not only on the state of the system, but on the structure within which that state is interpreted. This structure may include:

• measurement choices

• prior assumptions

• constraints on admissible states

• rules of inference

• institutional or procedural conditions

These are often treated as background or auxiliary. But in many cases, they are not optional. Without them, the system is not fully specified in a way that supports prediction or explanation.

Structured representation

We can make this explicit by distinguishing between:

• the state of a system (ψ)

• the contextual structure under which it is evaluated (μ)

Together, these form an extended representation:

Ψ = (ψ, μ)

Under this view, outcomes are determined not by ψ alone, but by the structured representation Ψ. The context μ is not merely additional information. It determines:

• which states are admissible

• which inferences are well defined

• which observations can occur

In this sense, context is not external to the model. It is part of what makes the model valid.

Conditional inference, structurally understood

Scientific reasoning is often described as conditional. We evaluate outcomes or probabilities given certain variables.

But there is a deeper layer.

The conditions themselves are not fixed. They are structured. The space of admissible outcomes, and the rules by which we condition, depend on the contextual structure within which the system is represented.

So inference is not just conditional. It is conditionally structured.

This helps explain why the same system, described in the same basic terms, can yield different results under different frameworks. The difference is not necessarily in the object, but in the structure under which it is evaluated.

Why this matters

This shift has several implications.

First, it explains why models that appear complete can fail. If relevant contextual structure is omitted, the representation is not merely simplified, but incomplete in a way that can change outcomes.

Second, it reframes disagreement. Apparent contradictions across domains may arise not because the underlying systems differ, but because the structures used to evaluate them differ and remain implicit.

Third, it suggests a different approach to unification. Instead of reducing all domains to a single ontology, we can align models by making their representational structures explicit and comparable.

A different picture of scientific knowledge

This does not mean that objects are unimportant, or that reality is subjective. The point is not to replace objects with context, but to recognise that objects are, in general, only scientifically meaningful when represented under appropriate structure.

The framework therefore allows us to describe the structured conditions under which things become operationally real, that is, observable, inferable, and admissible within a given representational setting.

In this sense, science is not only concerned with what exists, but with the conditions under which what exists becomes accessible to us in a stable and meaningful way.

Closing thought

If this is right, then one of the quiet assumptions of science, that specifying the object or system state ψ is sufficient to determine outcomes, requires explicit formulation. The framework characterises the conditions under which this assumption fails, namely when contextual structure alters admissibility or outcome distributions, and provides a formal basis for extending the representation.

Objects do not determine outcomes on their own.
Outcomes arise from objects under structure.

And it is this structure that science must learn to make explicit