A fallacy of abstraction, whereby when there is a discrepancy between a model and reality, it is reality (or its representation) that is manipulated to fit the model, rather than the model being adapted to reality.

This involves, for example, selectively choosing data or even falsifying it. In extreme cases, physical reality can even be manipulated to fit the theoretical model more closely.

The best way to explain this fallacy is to give a real-life example:

In this scenario, the first two options are both based on fallacious thinking, albeit in different ways:

1. Fitting the parcel to the car is akin to the Procrustean fallacy: reality is altered to fit the model – in this case, quite literally, the model of a car.
2. The second case looks like a motley mix of ignorance, wishful thinking and a reckless “it’ll turn out all right” attitude.

• “Procrustean bed”

In Greek mythology, “Procrustes” (Ancient Greek: Prokroústēs [Προκρούστης] = “the stretcher”) was another name for the giant Polypemon, a son of Poseidon.¹⁾

According to legend, Procrustes would force travellers to spend the night in his “iron bed”. If the bed proved to be too short for the guest, the visitor was summarily “made to fit” by having his legs chopped off. If the guest was too short, however, Procrustes would stretch them out to the correct length.

This was clearly not in the best interests of the (involuntary) guests: they would surely have preferred even an uncomfortable (because too short) bed – but if anything ought to have been adjusted in length, surely it would have been the bed rather than the guest.

The fundamental idea behind any model is that it serves as an abstraction of reality. Even though no model can be perfect, the relationship between the model and reality should always be clear: If there is a discrepancy, it is the model that must be adjusted, not reality.

The fallacy described here reverses this relationship: It is not the model that is adapted to reality, but reality that has to adapt to the model. In this respect, the approach resembles the macabre practice of Procrustes, who forced his guests to fit his bed – instead of the other way around.

The term is primarily used in academic discourse. However, that does not mean that similar phenomena cannot also be found in other fields, and indeed, there are many examples of this. The introductory example above is, unfortunately, not made up.

The classification of this fallacy within the categories on this website depends on which aspect one considers to be the most important:

• If one considers that the most important aspect is that an abstract model is regarded here as equivalent to, or even superior to, reality, then this should be labelled as some form of reification – perhaps even as what could be called “hyper-reification”, since the abstraction is given precedence over reality. Therefore, in this case, it should be classified as a “fallacy of abstraction”.

• Alternatively, one can also focus on the aspect of reversal that is taking place: The correct epistemic direction would be “reality → model” (i.e. reality determines what the model should look like). In the Procrustean fallacy, however, this becomes “model → reality”. If one considers this aspect to be the more important one, then it is primarily a fallacy of reversal.

While Procrustes’ “iron bed” does indeed describe a tangible object in which the “reversal” aspect plays a key role, most real-world instances of this fallacy involve abstract concepts that conflict with reality. For this reason, this fallacy is classified here as a “fallacy of abstraction”.

A scientific theory is, first and foremost, a model intended to explain certain phenomena in reality. As such a theory develops, situations often arise where the model and the phenomenon do not align. Ideally, the model should then be adapted to fit the phenomenon – a challenging process in which researchers may realise that months or even years of work were wasted due to being built on a flawed premise.

That can make it very tempting to simply manipulate the data so that it matches the “Procrustean bed” of the theory – following the motto: “If it doesn’t fit, force it.”

In the simplest scenario, one could simply discard data points that do not align with the theory, labelling them as “measurement errors” or “outliers” that are not worth further consideration (“epistemic trimming”). The next step then involves adjusting the methodology: A little data dredging for the analysis should suffice to make the model and data fit again. And ultimately, there have even been cases in which data and artefacts have been fabricated entirely.

One might be forgiven for writing off such practices as evidence of pseudoscience. However, the numerous recent scandals involving scientific misconduct demonstrate that this mindset is also prevalent within the scientific mainstream.

The Procrustean fallacy can also be understood as an aggravated form of other documented fallacies, the most notable of which is the so-called “confirmation bias”. This is a cognitive bias that describes the tendency to perceive information and arguments that seem to support one’s own opinion as more important and relevant than those that contradict it.

Closely related is a way of thinking referred to on this site as “coherentism”. In this paradigm, the internal consistency (“coherence”) of a conceptual framework (e.g., a worldview or a narrative) is overemphasized to such an extent that contradictions with reality are not even perceived as such in the first place.

Unlike the Procrustean fallacy, these two fallacies are essentially passive in that information that does not fit is simply filtered out or ignored. The issue here, however, is the active manipulation of information to make it consistent with the model.

Incidentally, the reverse can also be a fallacy if taken too far. In statistics, “overfitting” refers to the error of fitting a model too closely to the given data. This results in the model failing to capture the underlying pattern, meaning it can no longer be reliably applied to new observations.

A patient consults a specialist doctor regarding a specific health issue. The doctor has a preferred explanation for this type of condition: his theory is that it is primarily stress-related. During the consultation, the doctor steers the conversation towards potentially stressful situations at work or in the patient’s personal life. Other possible causes, such as diet, physical exertion or genetic predisposition, are, however, only mentioned in passing.

After a few turns, a suitable stressful situation is finally identified – for example, a conflict-ridden project at work or a tense family situation. The doctor is satisfied, as his theory that this condition is triggered by stress has been confirmed. The patient is also satisfied, as he has received a detailed assessment and may even have been prescribed a few days’ rest to recover from the “stress”.

The problem? It remains unclear whether stress was the actual cause, or at least the main cause, of the patient’s symptoms. The doctor did not conduct an unbiased anamnesis. Instead, he hastily settled on the hypothesis that “stress is the cause” and subsequently structured the consultation in such a way that this hypothesis was confirmed.

The distinction from other fallacies is subtle here: in the case of a mere confirmation bias, the doctor would passively look for evidence that confirms his theory and overlook any contrary indications. With coherentism, by contrast, the doctor would prioritise the internal consistency of his theory over its ability to truly reflect the patient’s complex situation, ignoring or downplaying contradictory facts. In the case of the Procrustes fallacy, the perception of reality is actively shaped so that it ultimately fits the model. In this example, the “Procrustes’ bed” corresponds to the theoretical framework that the doctor is determined to fit the patient into, and the anamnesis corresponds to the guest. Through leading questions, the patient’s testimony was then shaped (stretched out or trimmed) to make it fit the hypothesis.

However, this means that the underlying cause of the condition still remains undiagnosed. It is possible that the patient will indeed feel better after a few days’ rest. But perhaps they have also wasted valuable time, during which the illness could have progressed.

A mid-sized company has been using its own in-house customer management system for many years. This system consists of a tangle of spreadsheets, macros and various interfaces. While not particularly elegant, it effectively reflects the company’s business processes. To professionalise its internal infrastructure, the management has now decided to switch to a commercial CRM system.

Market research, however, reveals a dilemma: established systems follow a fixed internal logic with specific data structures, defined process steps, and pre-determined roles and responsibilities. The company must either invest substantial sums to adapt the software to its own workflows or restructure its existing processes so that they fit within the software’s framework. While the latter option may initially seem cheaper, it also incurs transition costs in the form of training, temporary productivity losses and, most likely, resistance from the workforce.

The second option is where the Procrustean fallacy lurks: The second option means that reality must adapt to the tool – not the other way around. The “bed” is the software’s data model and process logic, whilst the “guest” is the company’s actual workflows. Instead of adapting the software to fit the “shape” of the company, departments have to be reorganised, responsibilities shifted and established practices are abandoned until everything fits into the prescribed fields and workflows. The software remains unchanged, whilst the company is forced to adapt.

• Confirmation bias
• Overfitting (statistics)
• Reification