Choice (Microeconomic Foundations) – SAT – Evidence – Detection Limits

Back to Choice (Microeconomic Foundations) – SAT – Evidence

The Ten Cross-Scientific Detection-Limit Invariants

1. Sensitivity vs. Noise Floor

What “sensitivity” means in this field

In microeconomic choice, the “signal” is not a physical quantity but an expression of preference, valuation, or belief inferred from observed behavior (choices, prices paid, quantities chosen, response times, stated answers). Sensitivity therefore refers to the field’s ability to detect small differences in preference strength or tradeoff structure from noisy behavioral data.

The effective signal must be strong enough to stand out against behavioral variability, reporting error, situational fluctuation, and measurement imprecision.

Sources of noise

Across microeconomic settings, noise arises from multiple, structurally unavoidable sources:

These noise sources impose a baseline level of variability that cannot be eliminated, only reduced or modeled.

The detection boundary

A preference difference is detectable only if it produces behavioral regularities that exceed this noise floor. Below that threshold:

This is why many micro models rely on revealed preference inequalities or coarse choice patterns rather than attempting to recover exact utility levels.

Empirical manifestations of the limit

This detection limit appears concretely as:

In experimental settings, it shows up as high variance in repeated-choice tasks; in observational data, as instability of preference estimates under alternative controls.

Consequences for inference

Because of this limit:

Crucially, this is not a methodological failure but a structural detection constraint of the domain.

What lies beyond the limit

Below the sensitivity threshold lie:

These may exist, but they do not enter evidence without amplification (stronger incentives, repeated trials, sharper contrasts, or structural assumptions).

In Choice (Microeconomic Foundations), detection is limited by the field’s ability to separate true preference signals from unavoidable behavioral and measurement noise. Only preference differences large enough to generate stable, repeatable choice patterns rise above the noise floor and become empirically observable; finer structure remains inferentially inaccessible.

2. Resolution (Spatial, Temporal, Spectral, Angular)

What “resolution” means in this field

In microeconomic choice, resolution governs the field’s ability to separate distinct decision components—preferences, constraints, beliefs, tradeoffs, and timing—rather than merely detecting that a choice occurred.

There is no physical space or wavelength here. Instead, resolution operates over decision dimensions:

Resolution limits determine whether choice appears coarse and categorical or fine-grained and structured.

Sources of resolution limits

Resolution in choice is constrained by:

These limits are structural, not merely data limitations.

The resolution boundary

Below the field’s effective resolution:

At this boundary, distinct internal states map to identical observed choices.

Empirical manifestations of the limit

Resolution limits show up as:

Even with large datasets, these effects persist.

Consequences for inference

Because of resolution limits:

Resolution, not noise, sets the grain of microeconomic explanation.

What lies beyond the limit

Beyond observable resolution lie:

These may exist, but they are below the decision-resolution grain of the field.

In Choice (Microeconomic Foundations), detection is limited by the field’s ability to resolve distinct decision components from coarse, discrete, and bundled choice observations. Below this resolution, preferences, tradeoffs, and timing merge, forcing microeconomics to describe behavior at a coarser grain than agents’ internal decision processes.

3. Dynamic Range and Saturation

What “dynamic range” means in this field

In microeconomic choice, dynamic range refers to the field’s ability to simultaneously detect weak and strong preference signals within the same observational or experimental framework.

The “detector” here is the choice environment and measurement scheme: price variation, incentive magnitude, task difficulty, survey scale, or experimental payoff structure. Saturation occurs when choices become uninformative at extremes—either because differences are too small to matter or so large that behavior collapses to corners.

Sources of dynamic-range limits

Dynamic range in choice is constrained by:

These constraints bound how much preference variation can be expressed.

The saturation boundary

At the low end of dynamic range:

At the high end:

In both cases, the detector returns little information.

Empirical manifestations of the limit

Dynamic-range limits appear as:

Increasing variation often reveals one side of the range while obscuring the other.

Consequences for inference

Because of dynamic-range limits:

Dynamic range constrains what parts of the preference space are observable.

What lies beyond the limit

Beyond observable dynamic range lie:

These regions exist but are empirically silent.

In Choice (Microeconomic Foundations), detection is limited by the finite dynamic range of choice environments. Weak preferences vanish into noise, while strong preferences saturate behavior into corners, preventing simultaneous observation of the full preference spectrum within a single design.

4. Sampling Density, Coverage, and Missingness

What “sampling density and coverage” mean in this field

In microeconomic choice, sampling density and coverage determine which agents, decisions, and contexts ever enter the data at all. Detection is constrained not by how well choices are measured, but by which choices are observed, how often, and under what conditions.

The “sample” may be survey respondents, experimental subjects, transaction records, or observational datasets. Missingness arises whenever choices occur outside the observational frame.

Sources of sampling limits

Sampling constraints in choice arise from:

These gaps are structural, not random.

The coverage boundary

Below effective sampling coverage:

What is not sampled does not exist empirically.

Empirical manifestations of the limit

Sampling limits show up as:

Even perfect measurement cannot recover what was never observed.

Consequences for inference

Because of sampling limits:

Sampling defines the empirical universe of choice.

What lies beyond the limit

Beyond sampling coverage lie:

These choices may matter most, yet remain unseen.

In Choice (Microeconomic Foundations), detection is limited by who is sampled, which decisions are observed, and under what contexts. Sparse or biased coverage creates structural absences in the data, rendering entire regions of preference space empirically invisible regardless of measurement precision.

5. Channel Access, Penetration, and Occlusion

What “channel access” means in this field

In microeconomic choice, channel access refers to whether the analyst can observe the decision process or its determinants at all, not merely how accurately choices are measured. The “channel” is the pathway from an agent’s internal state (preferences, beliefs, constraints) to observable data.

Occlusion occurs when this pathway is blocked, indirect, or deliberately hidden, so that key components of choice never reach the observer.

Sources of channel occlusion

Channel access in choice is limited by:

These barriers are intrinsic, not technological.

The access boundary

Below effective channel access:

The phenomenon exists, but the observation path is closed.

Empirical manifestations of the limit

Channel occlusion appears as:

Data show what was chosen, not why.

Consequences for inference

Because of channel-access limits:

Choice models necessarily compress internal reality.

What lies beyond the limit

Beyond accessible channels lie:

These shape choice without being directly observable.

In Choice (Microeconomic Foundations), detection is limited by occlusion of the decision channel itself. Preferences, beliefs, and constraints are real but internal, reaching the observer only through indirect behavioral traces; where that channel is blocked or distorted, key determinants of choice remain empirically inaccessible.

6. Confounding, Interference, and Identifiability

What “confounding and identifiability” mean in this field

In microeconomic choice, this detection limit governs whether observed behavior can be uniquely attributed to a specific underlying cause—preferences, beliefs, constraints, or information—rather than merely detected as a behavioral outcome.

The issue is not whether choices are observed (they are), but whether the mapping from latent decision drivers to observed choice is one-to-one. Identifiability fails when multiple internal mechanisms generate indistinguishable choice patterns.

Sources of confounding and interference

Confounding in choice arises from structurally overlapping causal pathways:

These mechanisms interfere at the point of observable action.

The identifiability boundary

Below effective identifiability:

The data detect behavior, but not its unique cause.

Empirical manifestations of the limit

Identifiability limits appear as:

Identification often hinges on strong auxiliary assumptions.

Consequences for inference

Because of confounding and identifiability limits:

Microeconomic inference is therefore structurally underdetermined.

What lies beyond the limit

Beyond identifiability lie:

These distinctions may be meaningful but are empirically inaccessible without strong structure.

In Choice (Microeconomic Foundations), detection is limited not by observing behavior, but by confounding among latent decision drivers. Multiple combinations of preferences, beliefs, and constraints can generate identical choices, imposing identifiability limits that prevent unique causal attribution from choice data alone.

7. Calibration Drift and Definition Instability

What “calibration drift and definition instability” mean in this field

In microeconomic choice, this detection limit concerns whether observed choices remain comparable across time, contexts, instruments, and datasets. The issue is not whether behavior is measured correctly at a moment, but whether measurements retain a stable meaning as environments, elicitation methods, and definitions change.

The “calibration” here includes experimental designs, survey instruments, price scales, incentive schemes, task framing, and data-generation processes. Drift occurs when these evolve, intentionally or not.

Sources of instability

Calibration drift and definition instability in choice arise from:

These changes accumulate gradually and often invisibly.

The stability boundary

Below effective stability:

The signal exists, but its meaning moves.

Empirical manifestations of the limit

Instability appears as:

Micro data often lack a fixed reference frame.

Consequences for inference

Because of calibration and definition instability:

Choice data are locally valid but globally fragile.

What lies beyond the limit

Beyond stable calibration lie:

These are conceptually meaningful but empirically unstable.

In Choice (Microeconomic Foundations), detection is limited by calibration drift and instability of preference definitions. Measurements may be valid within a given instrument and context, but shifting elicitation methods and reference frames undermine stable comparison across time, datasets, and environments.

8. Rarity and Statistical Power

What “rarity and statistical power” mean in this field

In microeconomic choice, this detection limit concerns whether infrequent decisions or weak behavioral effects occur often enough to be empirically distinguishable from noise. The issue is not whether a choice pattern exists, but whether it appears with sufficient frequency or magnitude to be detected reliably.

Here, rarity can apply to types of agents, types of decisions, or types of contexts. Statistical power governs whether observed data can reject randomness or null models in favor of meaningful structure.

Sources of rarity and low power

Rarity in choice arises from several structural features:

Even large datasets may contain few relevant observations.

The power boundary

Below effective statistical power:

Absence of evidence reflects insufficient observation, not absence of structure.

Empirical manifestations of the limit

Rarity and power limits appear as:

Choice data often privilege common behavior.

Consequences for inference

Because of rarity and power limits:

Inference is biased toward frequent, average decisions.

What lies beyond the limit

Beyond observable power lie:

These matter disproportionately but are empirically thin.

In Choice (Microeconomic Foundations), detection is limited by rarity of relevant decisions and insufficient statistical power. When choices or effects occur infrequently or weakly, their absence from data reflects limited observation rather than nonexistence, constraining inference about tails and subtle behavioral structure.

9. Measurement Back-Action and Disturbance

What “measurement back-action” means in this field

In microeconomic choice, measurement back-action occurs when the act of observing, eliciting, or incentivizing a choice alters the choice itself. Below a certain scale or subtlety, the measurement procedure dominates the phenomenon it seeks to detect.

The “measurement” includes surveys, experiments, incentives, framing, observation, and data collection itself. Disturbance arises because agents respond to being measured.

Sources of measurement disturbance

Back-action in choice arises from structurally unavoidable mechanisms:

Measurement reshapes the decision environment.

The disturbance boundary

Below effective non-disturbance:

The signal exists, but the probe overwhelms it.

Empirical manifestations of the limit

Back-action appears as:

Measurement effects are often larger than the effects of interest.

Consequences for inference

Because of measurement back-action:

Choice data are inseparable from their measurement context.

What lies beyond the limit

Beyond non-disturbing measurement lie:

These cannot be accessed without reshaping them.

In Choice (Microeconomic Foundations), detection is limited by measurement back-action. The act of eliciting or observing choice alters the decision environment itself, so beyond a certain subtlety, measurement reshapes what it seeks to reveal, imposing intrinsic limits on preference observability.

10. Computational and Algorithmic Tractability

What “computational tractability” means in this field

In microeconomic choice, computational and algorithmic tractability limits whether latent preferences, beliefs, and decision rules can be recovered, estimated, or simulated within feasible computational bounds. The constraint is not empirical detection, but whether the inference problem itself is solvable with available algorithms and resources.

Here the “instrument” is the estimation and optimization machinery: likelihood evaluation, search over preference spaces, dynamic programming, equilibrium computation, and simulation-based inference.

Sources of computational intractability

Tractability limits in choice arise from structural features of decision models:

These limits arise even with perfect data.

The tractability boundary

Below effective tractability:

The phenomenon exists, but cannot be computed.

Empirical manifestations of the limit

Computational limits appear as:

What is estimated reflects tractability constraints as much as behavior.

Consequences for inference

Because of computational limits:

Inference is bounded by algorithmic reach.

What lies beyond the limit

Beyond tractability lie:

These may exist but cannot be computed at scale.

In Choice (Microeconomic Foundations), detection is limited by computational and algorithmic tractability. Even when preferences and decision rules are conceptually well-defined and empirically detectable, they may lie beyond feasible computation, forcing inference to operate within simplified, solvable representations rather than the full decision space.