Concepts · Evidence and Explanation

Different Causes, Same Outcome

Different causal pathways can converge on the same observed outcome. The outcome itself establishes neither that the underlying pathways are the same nor that they are different.

By Yona Ole Lobulu ·

Concept5 min readD2.7

Topic
Evidence and Explanation
Read first
One piece should be read before this one
Reading time
About 5 minutes of reading
Difficulty
Advanced reading: this page assumes a fair amount of earlier reading

The question

What does an observed outcome establish about the causal pathway that produced it?

Definition

Different causal pathways can converge on the same observed outcome, but the outcome itself establishes neither that the underlying pathways are the same nor that they are different.

Two people decline the same social invitation.

From the outside, the behaviour looks the same.

But the same non-attendance could be compatible with several explanations: anticipated social threat, exhaustion, competing demands, environmental constraint, or combinations among them. The behaviour alone does not tell us which, if any, of these explanations applies.

This points to a basic problem in causal explanation:

Different causal pathways can converge on the same observed outcome.

Different pathways can converge on the same outcome

A simple causal structure looks like this:

Pathway A → Outcome Y

Pathway B → Outcome Y

Pathway C → Outcome Y

Different routes can arrive at the same or similar endpoint.

Developmental science often describes this many-to-one pattern as equifinality: different developmental pathways can converge on the same or similar outcome.

The underlying causal structure is more general. Similar observed outcomes in behaviour, cognition, bodily functioning or social interaction can also be compatible with different causal pathways.

Those pathways may differ in their causes, conditions, constraints, intermediate processes, mechanisms or prior histories.

Two outcomes can therefore look the same without having arrived there in the same way.

But that does not mean they must have arrived differently.

The outcome itself does not settle the question.

The outcome does not uniquely reveal its pathway

Suppose all we observe is Y.

If pathways A, B and C are each capable of producing Y, observing Y alone does not uniquely identify which pathway occurred.

In this sense, the causal explanation is underdetermined by the outcome alone.

The endpoint still contains information. It may rule out some explanations, make others more plausible, or become more informative when combined with other evidence. What it does not provide is a complete record of the causal history that produced it.

This extends the distinction established in Correlation, Prediction, Causation and Mechanism.

Observing an outcome is not the same as identifying its cause. Identifying a cause is not the same as identifying the mechanism through which its influence occurred.

D2.7 adds another constraint:

Even cases with the same outcome may have arrived there through different causal routes.

The correct inference is therefore not:

same outcome, same cause.

Nor is it:

same outcome, different cause.

It is:

Outcome similarity leaves causal similarity unresolved.

Classification is not the same as explanation

Science frequently groups cases according to shared features.

Two people may:

perform the same behaviour;

report the same symptom;

receive the same score;

meet the same diagnostic criteria;

show the same recurring pattern.

Such classifications can be scientifically useful for describing, comparing and studying cases.

But classification and causal explanation answer different questions.

Classification asks:

What relevant feature do these cases share?

Causal explanation asks:

How was that feature produced?

The same exam score, for example, can be compatible with different combinations of knowledge, strategy, attention or error.

Likewise, two people can meet the same diagnostic criteria without the classification itself establishing that the causal pathways producing their difficulties are identical.

A shared category tells us something real about the dimension on which cases were grouped.

It does not automatically reveal a shared mechanism.

A shared classification does not automatically imply a shared causal explanation.

Different pathways can still share important components

Heterogeneity in causal pathways does not mean every case is produced by a completely unique causal system.

In a deliberately simplified schematic:

X + M + C₁ → Y

X + M + C₂ → Y

The pathways share X and M but differ in C₁ and C₂. Both still produce Y.

One case may therefore differ from another in only part of its causal configuration.

Two people can share some causes, conditions, mechanisms or developmental influences while differing in other relevant factors.

This matters because the alternative to one universal pathway is not complete individual uniqueness.

Different pathways can still share important components.

How much they share is an empirical question.

Multiple pathways do not mean randomness

If more than one route can produce an outcome, it may sound as though causal explanation becomes arbitrary.

It does not.

Different pathways can themselves be structured. Some may recur under particular environmental conditions. Others may depend on developmental history, bodily capacity, learned expectations, social conditions or particular combinations of causes and constraints.

More than one route to an outcome therefore does not mean that anything can produce anything.

Nor does it mean that general patterns disappear.

Researchers can identify recurring pathway components even when no single causal route applies to every case. Developmental research on equifinality provides one clear example: variation in pathways can itself be patterned rather than random.

General structure and causal heterogeneity can coexist.

What the outcome cannot tell us by itself

Recognising that multiple pathways are possible creates another risk: inventing individual explanations.

If the same behaviour could be compatible with threat, fatigue, constraint or competing incentives, it can be tempting to observe one person's behaviour and select whichever explanation seems most plausible.

But possibility is not identification.

Knowing that pathways A, B and C can all produce Y does not establish which pathway produced this particular Y.

That requires additional evidence.

Causal heterogeneity therefore places a limit on inference from the outcome alone. It does not provide a shortcut to explaining an individual case.

Several hypotheses may be compatible with the same observation. Distinguishing among them requires evidence that can discriminate between those hypotheses.

Same outcome does not imply same explanation

The principle can now be stated precisely.

The same observed outcome can be produced through different causal pathways.

But similar outcomes do not prove that their pathways differ. Cases may share substantial causal structure, differ only in some conditions or mechanisms, or arise through substantially different histories.

The observation alone does not decide among those possibilities.

So:

Same behaviour does not establish the same cause.

Same score does not establish the same process.

Same classification does not establish the same mechanism.

The causal question remains open until additional evidence can distinguish among the possible explanations.

Outcome similarity leaves causal similarity unresolved.

The outcome shows us what occurred. By itself, it does not uniquely reveal how that outcome was produced.

Sources and research record3 sources, with findings, strengths and limitations as entered

References

3 sources this piece rests on, as entered in the Library.

  1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

    Theoretical article · Development and Psychopathology, 8(4) · 597–600

    The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

    doi:10.1017/S0954579400007318

  2. Maung, H. H. (2016) Diagnosis and causal explanation in psychiatry

    Theoretical article · Studies in History and Philosophy of Science Part C, 60 · 15–24

    Examines causal heterogeneity within psychiatric diagnostic categories, arguing that a diagnosis can be informative without identifying one specific underlying causal process.

    doi:10.1016/j.shpsc.2016.09.003

  3. Bolger, N. et al. (2019) Causal processes in psychology are heterogeneous

    Methodological article · Journal of Experimental Psychology: General, 148(4) · 601–618

    Argues that heterogeneity in causal effects is common in psychology, can be modelled explicitly, and is scientifically informative rather than mere noise around an average.

    doi:10.1037/xge0000558

Further reading

Behind this page

The claims this concept makes, the evidence behind them, and the limits it accepts.

Evidence status

High confidence

Strongly supported, though resting on synthesis or principle rather than a single decisive body of evidence.

Claims

  1. Different causal pathways can converge on the same or a similar observed outcome

    High confidence

    What this does not assert: The claim concerns what is possible in causal structure, not what occurred in any particular case.

    1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

      Theoretical article · Development and Psychopathology, 8(4) · 597–600

      The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

      doi:10.1017/S0954579400007318

  2. Multiple pathways to an outcome does not mean causal randomness

    High confidence

    What this does not assert: Variation in pathways can itself be patterned, and recurring pathway components can be identified.

    1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

      Theoretical article · Development and Psychopathology, 8(4) · 597–600

      The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

      doi:10.1017/S0954579400007318

    2. Bolger, N. et al. (2019) Causal processes in psychology are heterogeneous

      Methodological article · Journal of Experimental Psychology: General, 148(4) · 601–618

      Argues that heterogeneity in causal effects is common in psychology, can be modelled explicitly, and is scientifically informative rather than mere noise around an average.

      doi:10.1037/xge0000558

  3. Pathway heterogeneity is not the same as heterogeneity in causal effects

    Canonical inference

    What this does not assert: Research on effect heterogeneity supports the general point that heterogeneity can be modelled; it does not make the two notions equivalent.

    1. Bolger, N. et al. (2019) Causal processes in psychology are heterogeneous

      Methodological article · Journal of Experimental Psychology: General, 148(4) · 601–618

      Argues that heterogeneity in causal effects is common in psychology, can be modelled explicitly, and is scientifically informative rather than mere noise around an average.

      doi:10.1037/xge0000558

  4. Knowing which pathways could produce an outcome does not identify which one did

    Canonical synthesis

    What this does not assert: Distinguishing among compatible hypotheses requires evidence able to discriminate between them.

  5. Outcome similarity leaves causal similarity unresolved

    Canonical synthesis

    What this does not assert: The Library's compression of the principle: the outcome shows what occurred, not how it was produced.

  6. Developmental science calls the many-to-one pattern equifinality

    High confidence

    What this does not assert: Equifinality is a developmental-science term; the broader causal structure described here is more general than that literature's usage.

    1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

      Theoretical article · Development and Psychopathology, 8(4) · 597–600

      The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

      doi:10.1017/S0954579400007318

  7. Observing an outcome does not uniquely identify which pathway produced it

    Canonical inference

    What this does not assert: The endpoint can still rule out explanations and become informative when combined with other evidence.

    1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

      Theoretical article · Development and Psychopathology, 8(4) · 597–600

      The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

      doi:10.1017/S0954579400007318

  8. Outcome similarity does not establish that the causal pathways were the same

    Canonical inference

    What this does not assert: Nor does it establish that they differed; the question is left open.

  9. Outcome similarity does not establish that the causal pathways differed

    Canonical inference

    What this does not assert: Cases may share substantial causal structure and differ only in part of it.

  10. A shared classification does not automatically imply a shared causal explanation

    Established

    What this does not assert: Classification answers a descriptive question about a shared feature; it can be informative without identifying one causal process.

    1. Maung, H. H. (2016) Diagnosis and causal explanation in psychiatry

      Theoretical article · Studies in History and Philosophy of Science Part C, 60 · 15–24

      Examines causal heterogeneity within psychiatric diagnostic categories, arguing that a diagnosis can be informative without identifying one specific underlying causal process.

      doi:10.1016/j.shpsc.2016.09.003

  11. Two people can meet the same diagnostic criteria without identical causal pathways

    Established

    What this does not assert: This concerns causal heterogeneity within a category, not the clinical or scientific value of the category itself.

    1. Maung, H. H. (2016) Diagnosis and causal explanation in psychiatry

      Theoretical article · Studies in History and Philosophy of Science Part C, 60 · 15–24

      Examines causal heterogeneity within psychiatric diagnostic categories, arguing that a diagnosis can be informative without identifying one specific underlying causal process.

      doi:10.1016/j.shpsc.2016.09.003

  12. Different pathways can still share important components

    Canonical inference

    What this does not assert: How much any two cases share is an empirical question, not something the outcome settles.

  13. Causal heterogeneity does not imply that every case is causally unique

    Canonical inference

    What this does not assert: The alternative to one universal pathway is partial overlap, not complete individual uniqueness.

Sources

  1. Cicchetti, D., Rogosch, F. A. (1996) Equifinality and multifinality in developmental psychopathology

    Theoretical article · Development and Psychopathology, 8(4) · 597–600

    The foundational developmental statement of multifinality: similar developmental beginnings can lead to different developmental outcomes.

    doi:10.1017/S0954579400007318

  2. Maung, H. H. (2016) Diagnosis and causal explanation in psychiatry

    Theoretical article · Studies in History and Philosophy of Science Part C, 60 · 15–24

    Examines causal heterogeneity within psychiatric diagnostic categories, arguing that a diagnosis can be informative without identifying one specific underlying causal process.

    doi:10.1016/j.shpsc.2016.09.003

  3. Bolger, N. et al. (2019) Causal processes in psychology are heterogeneous

    Methodological article · Journal of Experimental Psychology: General, 148(4) · 601–618

    Argues that heterogeneity in causal effects is common in psychology, can be modelled explicitly, and is scientifically informative rather than mere noise around an average.

    doi:10.1037/xge0000558

Where to go from here

Next published piece

Same Cause, Different Outcomes

A cause can be real without having a uniform consequence. The same causal influence can produce different effects under different conditions.

Continue through the Library →See where this sits in the graph →

Back to the Library →