Nousophy: Beyond the Four Letters

In 2017 I took a personality test and received four letters: ENTP. Within roughly a year, another result gave me INTP. Neither felt entirely wrong. I could recognise myself in both, which left me wondering what had changed: my behaviour, the circumstances I had in mind while answering, my understanding of myself, or the way a difference in responses became a difference in type.

The two results could not settle that question. A changed label does not tell us how much changed underneath it, and a stable label would not establish that everything underneath had stayed the same. I became interested in what the result allowed me to see and what I would need to examine separately.

Official feedback can include preference clarity, discussion of a best-fit type, and, in Step II, more detailed facets. Preference clarity describes how consistently someone answered towards a preference; it is not a measurement of how much of a trait they possess. The difficulty I wanted to pursue was the distance between recognising a description and understanding why a person behaves as they do.

Running a large community in which many people have taken typology tests, I have seen people respond to a changed result by dismissing the test or choosing whichever type feels closer. I understand the appeal of either response. A label can supply language for something a person has struggled to explain, and once that recognition has occurred, investigating the scoring may seem beside the point. For me, the disagreement between two plausible descriptions became a reason to keep looking.

A category compresses information. That can make it useful: a short description is easier to remember, communicate and compare than a long record of someone's responses. The cost depends on what has been compressed and what we subsequently ask the category to explain. Two people can receive the same result while differing in the consistency of their answers or the situations to which those answers apply. A type name alone cannot recover those differences.

Over the years since those results, I have examined the questions, scoring decisions and assumptions behind typology systems. For the past few years I have also been developing my own assessment model, drawing on Jungian function theory and the ideas behind MBTI. I want to retain differences within a profile and examine combinations of tendencies, including combinations that do not fit comfortably into a single type.

The model groups and scores answers across proposed dimensions, with attention to preference, consistency, and tensions between responses. It distinguishes strong clustering from a mild inclination within its scoring rules. Those rules describe how I have organised the answers. Calling one cluster an analytical tendency may be useful shorthand, but demonstrating a distinct cognitive process requires evidence beyond the name I have given it.

Inconsistency interests me because a single answer can conceal a condition. Someone might enjoy disagreement when discussing an abstract problem and avoid it when a relationship feels at risk. Asking about both could reveal something a broad question misses. Conflicting answers could also arise from an ambiguous item, a change in mood, an inaccurate recollection, or a distinction that exists mainly in the questionnaire.

Self-report places another limit on what can be learned this way. People answer from the situations they remember, the self-description they currently accept, and sometimes the person they would prefer to be. More questions may help distinguish some of these possibilities, although the resulting detail still comes from the same source. A question about behaviour under pressure records someone's report of that behaviour. It does not, by itself, show how they will act when the pressure arrives.

Consider a hypothetical profile in which someone reports sustained analytical effort, extensive anticipation of future problems, and withdrawal when emotional uncertainty increases. I might hypothesise that analysis gives this person a relatively manageable task when a conversation feels difficult, so they continue solving the problem while postponing the conversation. That would connect behaviours that can look unrelated. The answers alone cannot tell me whether that account is right. Withdrawal could reflect exhaustion, a reasonable judgement about a particular relationship, or simply a need for time before responding.

To test the proposed connection, I would need to state in advance what the profile predicts, in which circumstances, and what observation would count against it. For example, does the response pattern anticipate later postponement of difficult conversations, beyond what we could predict from the person's existing habit of postponing them or from their current workload? Later observations would need to be collected separately from the answers used to construct the profile. If the model predicts withdrawal largely because its questions already ask whether someone withdraws, it has rearranged a description without demonstrating much predictive gain.

A useful test would also allow the simpler account to win. If workload predicts the later behaviour as well as the proposed configuration does, I would have less reason to invoke a combination of underlying processes. Revising a model after a failed prediction may improve it, but the revised explanation would then need to face further observations. Otherwise each failure supplies another exception and the model never has to give anything up.

The same caution applies to a relational profile. Reports of frequent accommodation, difficulty refusing requests and reliance on approval could suggest a pattern worth examining, including whether resentment follows commitments the person did not want to make. A person with little freedom to refuse may give similar answers for very different reasons. Describing their conduct as poor integration could turn a constraint in their circumstances into a defect in their personality. The conditions under which an answer was given belong in the interpretation.

I am also interested in whether combinations of responses correspond to differences in the tone of someone's internal narration: how they anticipate, criticise, reassure or explain themselves. I would compare those predictions with people's own accounts, collected independently and examined for differences as well as matches.

The ambition is to understand configurations that change with conditions. I want to examine when a capacity helps, when it becomes difficult to use, and how it interacts with other tendencies.

By integration, I mean the possibility of using capacities with more awareness of their conditions and consequences. That remains an aim of the work. Showing that a questionnaire describes a pattern would not show that acting on its interpretation improves someone's life. Such a claim would require its own evidence about the proposed action and its effects. There is also room for a finding less flattering to the model: what looks like an internal weakness may be better addressed by changing an expectation, a demand, or the situation in which the person is being assessed.

I still want to build something that helps people notice patterns a familiar identity has made easy to overlook. A profile should make it easier to ask a precise question, compare an interpretation with experience, and revise it when it fails. For now, the scoring offers an organised description and a set of proposed relationships to investigate. The work beyond that is to find out which of those relationships hold, how far they hold, and which I will need to abandon.

Further reading: The Myers & Briggs Foundation: My MBTI Results; The Myers-Briggs Company: MBTI facts.