Parsimony in psychology is the principle that the simplest explanation for a behavior or mental process is the one you should prefer. When two theories explain the same set of facts equally well, the one that makes fewer assumptions, proposes fewer unseen mechanisms, and applies more broadly wins. It’s the psychological version of a rule you may already know: Occam’s Razor.
But parsimony isn’t just a philosophical nicety. It shapes how researchers build theories, how clinicians diagnose mental health conditions, and how the field decides which ideas deserve to survive.
The Formal Principle
The American Psychological Association defines the law of parsimony as the principle that the simplest explanation of an event or observation is preferred. “Simplest” here has specific criteria. A parsimonious explanation makes the smallest number of unsupported assumptions, proposes the fewest entities, and relies on the fewest things that can’t be directly observed. A useful formulation from the research literature puts it this way: where two theories account for the same facts, prefer the one that is briefer, makes assumptions you can easily drop, refers to things you can actually observe, and has the greatest generality.
This isn’t the same as saying reality is always simple. Parsimony is a tiebreaker. It tells you what to prefer when two explanations are otherwise equal in their ability to account for the evidence. If a more complex theory genuinely explains something a simpler one cannot, complexity is justified. The key word is “no reason to do otherwise.” Adding complexity should earn its keep.
Morgan’s Canon and Animal Behavior
Parsimony first became a formal guideline in psychology through the work of C. Lloyd Morgan, widely credited as the father of comparative psychology. In the late 1800s, scientists studying animal behavior had a habit of attributing rich mental lives to animals based on limited evidence. A dog that opened a gate latch might be described as reasoning through the problem, when trial-and-error learning could explain the same behavior.
Morgan’s Canon states that you should not interpret an animal’s action as the result of a higher mental process if it can be explained by a lower one. If simple learning accounts for the gate trick, don’t invoke abstract reasoning. This was parsimony applied directly to interpretation: don’t pile on mental complexity when a simpler mechanism fits the data. The canon has been debated and sometimes misapplied over the decades, but it established an enduring norm in psychology. Before proposing that something sophisticated is happening in the mind, rule out the straightforward explanations first.
How Parsimony Shapes Theory
In research, parsimony acts as a filter for competing theories. When psychologists propose models of how people think, feel, or behave, those models inevitably differ in complexity. One might introduce three new constructs to explain a pattern of results. Another might explain the same pattern with one. If both handle the existing evidence equally well, parsimony favors the leaner model.
A concrete example comes from research on cognitive biases. Psychologists have cataloged dozens of distinct biases, each with its own theoretical explanation. The bias blind spot (the tendency to see bias in others but not yourself) is sometimes explained by a motive for superiority. The hostile media bias (where partisans on both sides of an issue see the same news coverage as biased against them) gets its own separate explanation involving group identity and emotional investment. A more parsimonious framework, proposed in recent research, argues that many of these biases stem from a single, shared mechanism: people believe their own perceptions are correct, and they process new information in ways that stay consistent with that belief. If you assume your own view is accurate, then anyone who disagrees, whether a person or a news outlet, must be the biased one. You don’t need a separate motivational theory for each bias. One general process, belief-consistent information processing, covers a wide range of them.
This is parsimony in action. Rather than maintaining a dozen independent explanations for a dozen biases, a single framework with fewer assumptions does the same work. It also has greater generality, applying across contexts and populations rather than being tailored to one narrow phenomenon.
Parsimony in Clinical Diagnosis
The principle also matters in clinical settings, where it influences how mental health professionals classify and assess conditions. The current diagnostic system for anxiety alone contains 11 distinct categories, including generalized anxiety, social anxiety, panic disorder, agoraphobia, specific phobias, separation anxiety, selective mutism, PTSD, and OCD. A thorough assessment requires clinicians to evaluate over fifty symptoms, check symptom duration and impairment, and work through differential diagnosis separately for every category.
Recent research tested whether a simpler system could do just as well. A transdiagnostic dimensional model reduced all of that to a maximum of four dimensions, and the researchers found it was no worse at predicting outcomes than the full DSM classification. The four-dimension system could significantly cut the time required to assess anxiety pathology without losing predictive power. In fact, two of the proposed dimensions, intensity and avoidance, turned out to be so highly correlated that they were essentially measuring the same thing, suggesting the system could be simplified even further.
This is diagnostic parsimony: if a simpler classification captures the same clinical information as a complex one, the simpler version is better for everyone. Clinicians spend less time on assessment, patients get a more coherent treatment plan, and the overall system becomes easier to use reliably. Every time you add a new construct to a model, you increase the burden of assessment while reducing parsimony. That addition needs to prove it’s worth the cost.
Parsimony and Scientific Rigor
There’s a deeper reason parsimony matters beyond convenience. Simpler theories are easier to test and, critically, easier to disprove. The philosopher Karl Popper argued that good scientific theories should be falsifiable, meaning there should be possible observations that could prove them wrong. A theory loaded with extra assumptions and escape hatches is harder to falsify because each assumption can be adjusted to accommodate contradictory evidence. Every time you add an ad hoc explanation to rescue a theory from disconfirming data, you make the theory less parsimonious and, in Popper’s framework, less scientifically useful.
A parsimonious theory puts itself at risk. It says, “Here is a simple claim, and here is what would prove it wrong.” That vulnerability is a feature, not a bug. Theories that survive repeated opportunities to be disproven earn more credibility precisely because they didn’t need layers of protective assumptions to stay standing.
Where Parsimony Has Limits
Parsimony is a guide, not a guarantee. Human behavior is genuinely complex, and sometimes the simple explanation is wrong. A child’s disruptive behavior in school could reflect boredom (a simple explanation), but it could also reflect an anxiety disorder, a learning disability, and a difficult home environment all interacting at once. Forcing a single parsimonious explanation onto a situation that truly involves multiple causes leads to incomplete understanding and, in clinical contexts, inadequate treatment.
The principle works best as a starting point. Begin with the simplest adequate explanation, then add complexity only when the evidence demands it. If a single diagnosis accounts for all of a patient’s symptoms, start there. If it doesn’t, additional diagnoses are warranted. If one theoretical mechanism explains a set of experimental results, don’t propose three. But if one mechanism leaves patterns unexplained, the more complex model may be the right one. Parsimony tells you not to multiply explanations beyond necessity. It doesn’t tell you to ignore real complexity when it exists.