English · Русский · Українська
In 90 seconds
Someone keeps asking what will happen several years from now. We can hear the questions. But we do not yet know why they ask them.
One possible explanation is a particular way of organizing their experience of time. Such a proposed internal process is called latent because it cannot be observed directly. In Before We Build it is an idea to test. Assigning a type does not mean discovering a process.
Observation and explanation are different steps
Several things need to be distinguished. The whole person cannot be reduced to any of them.
- Observation and description: what someone said or did, and what recurs.
- Type hypothesis: a proposal that the recurrence fits a particular way of perceiving and organizing experience.
- Process hypothesis: a proposal about internal activity that might explain the recurrence.
- Natural-predisposition hypothesis: a proposal that this tendency has a natural basis.
Each further step needs separate evidence. Claiming a causal mechanism is an even stronger inference: fitting the observations is not enough.
Example: questions before a move
In several discussions about moving, someone asks to compare long-term scenarios. This is an observation. “They often consider future consequences” describes a recurrence, or pattern.
Interpreting this pattern through Temporistics proposes a type hypothesis. Proposing a particular way of organizing temporal experience is a latent-process hypothesis. None of these steps alone proves why the person acted that way.
What observations to collect
Use several indicators, rather than one memorable event. Consider behaviour across repeated tasks, words, timing, choices under uncertainty, and revisions to decisions. Compare what people say about themselves with their conduct.
Define the indicators in advance, before learning the outcome you want to predict. Otherwise it is easy to select only what fits an outcome you already know.
What the hypothesis needs to explain
A useful process hypothesis must explain why several observable indicators vary together. It must be distinguishable from other possible explanations and tested for whether it improves prediction in a specific domain.
Mappings between typological positions and processes remain unvalidated project hypotheses. A proposed natural basis for a tendency also needs separate evidence. It does not establish innateness.
What else could produce the same result
Skills, knowledge, incentives, social roles, stress, culture, and measurement errors may produce the same set of indicators. Even several different models of internal processes may fit the same observations equally well.
What is not enough for a conclusion
A convincing story does not prove a cause. A description of a type pattern is not the proposed latent process or variable itself. One striking case does not show that a characteristic is stable, applies everywhere, or establishes compatibility.
Questions for conversation
- What exactly did we observe?
- Are we describing a recurrence, proposing a type, or proposing an explanatory process?
- What other explanation predicts the same behaviour?
- What do we need to observe next to distinguish these explanations?
How to prepare a test
Before testing the model, write down what you will observe, the explanations you will compare with the hypothesis, and the outcome you want to predict. Specify which indicators should agree and which should differ, as well as the measurement error you expect. Read more about compatibility measurement methods.
Where to go next
Read about the limits of inference, then the reinterpretation of typologies. For explaining such ideas without losing accuracy, see the scientific narrative page.