Pervin asked: Is (Big5) Trait theory sufficient for Personality?

• Does Big5 Trait give a total theory of motive, thought, emotion, and all? No.
  
• Is Big5 validly predictive? At most modestly so.
  
• Are the Big5 dimensions a consensus set? No.
  
• Is the idea of Trait a consensus idea? No.

• confuses description with motivation,
  
• is more descriptive than explanatory
  (Chomsky said: description captures the data, explanation captures the inner or generating process)
  
• is limited in 4 ways, namely, it is:
  
  • 1) purely descriptive,
    
  • 2) nomothetic,
    
  • 3) psychometric,
    
  • 4) empirical/atheoretical

• 1) purely descriptive doesn't show what makes it tick, no internal process is revealed.

• 2) nomothetic (individual differences oriented approach) ignores everything within-individual.
  Again inside ≠ outside.
  Intrasubject emotion-tracking designs show lots of interesting stuff (Epstein 1983)

• 3) psychometric: "For some purposes, an atheoretical psychometric, empirical hodgepodge scale like Neuroticism is obviously useful. For others, it has little more to offer than a label." It might be useful to know how a person effectively handles their neuroticism, or not, and the effects thereof, which the mere label glosses over.

• 4) empirical/atheoretical: cognizers have process, and goals, and they process information; understanding them requires going a bit deeper and seeing their flexibility and variability: How do they think, feel, and operate? Unconscious and qualia-qualified processing under personality variation would enhance Big5's reach. Unconscious processing implies limitations in Big5 self-report data. Empirical similarities may be superficial: whale vs shark or mouse.

What stuck out to me

From Trait Theory falls out the idea of "IS", which in statistics is the idea of population statistics: +A vs -A are two subpopulations, are their means or variances the same or different using statistical tests. The logic here is has to do with the identity of the element, which in grammar uses the verb TO BE as in "This IS an A." or in predicate logic the predicate A(this) or the set theoretic assertion that this ∈ A., which are all equivalent, intertranslateable versions of the mental model of things being in one category or another.

Unfortunately for trait theory, things are more complicated than their category, and knowing even many of the categories which apply to any given one or thing or instance or element or subject or person, may not tell you everything you might want to know about it.

On the other hand, incuriosity is prevalent if not universal. "My ignorance is exceeded only by my lack of curiosity" is not just funny but recognizeable as characterizing a lot of the world, or at least my relationships with many. It's a perfect self-reflective universe: we blockheads have our categories, so we know plenty now, so we're done, quit bothering us.

Incuriosity is not an argument, but rather a choice, indeed a self-categorizing choice, which reveals a lot about a person's nature, a somewhat interesting data point, also disqualifies them participating in the conversation.

Knowing more can be desireable practically and scientifically, to understand what things are, are made of, their inputs and outputs, their internal activities resources and capabilities, and their external interactions both felicitous/effective and ineffective for it and for interactants. One might want to understand so as to better teach one's students or children, or improve this world. There is no limit to desire for knowledge.

Given interest in the inner workings of an information processor with modulated reasoning pathways, science could put the organism on the spot, provide it with inputs with certain logical characteristics, watch its behavior and infer reasoning pathways that reconcile the logic of stimulus and response. We externalize the problem and solutions, and examine behavior, inferring internal reasoning from problem and input data to solution and behavior.

Or do we?

A chess program could develop a mental model of the processing of a chess player by playing them enough times to see "how they think", perhaps. The domain and ruleset are small, external considerations can be ignored, reasoning can perhaps be characterized using search depth, working memory size, speed, previous-pattern matching memory-space size and retrieval speed and accuracy and generalizeability. Okay.

Many concrete versions of tautologically equivalent models may be generated by the chess program, and math and logic might reduce them to some simplest form such as a Kolmogorov information metric or model, imaginably (I speculate).

In domains of emotion, diplomacy, the learning of social intelligence, aids to mature behavior, etc., what would be the analog of the chess program building a mental model of people? of its domain, ruleset, processing pathways (can they even be called "reasoning"?) ? What simplest form of metric or model would capture the maximum-simple essence from interactions with a person to represent their qualities, capabilities, predilections in such domains? Might one (how) go from such a model to recommendations for teaching or learning, which are effective?

A set of traits might indeed help to do it, if they were translated into a process domain. How would that be validated? If predilection were inferred from (in)capability, that would be invalid.

In short, I don't see how population statistics gives us an easy path to decoding the circuitry of a computational mind, although without some kind of experimental and statistical reasoning noone would believe even block diagrams or flow charts or blueprints from God.

Time for a thought experiment: the Nomological Network for a minimal process step such as Choice or Decision. What are its elements? +A and -A are, say, the choices. Inputs such as motive, capability, perception, assessment-system, externally provided information, and internal settings or memories are then MODULATING variables which influence the choice of +A or -A. Enough experiments manipulating the modulating input variables might tease them apart (with manipulation checks of course) to determine the form and parameters of an imaginary internal equation that takes the inputs to the Choice output most predictively. Perhaps there are experiments or statistics that would bear on the question, given a good model, how do we know it is the best? (A Kolmogorov criterion may help: the shortest program to generate certain output is the complexity of the process.)

Generalizing, every bit in a reasoning system might have its own pilot study of a Choice or Decision identical to the above, and perhaps a larger architecture might be justified by a larger experiment combining these, assuming:

• discriminant validity of each bit from each other,
• reliability of each bit with itself,
• convergent validity of redundant-alternative inputs driving an individual bit,

(All this while minimizing noise, bringing manipulation causally close to construct, confirming attention, wakefulness, etc. in the subject, and other experimental validity checklist items.)

Then a "good" model would predict not just training data but test data, hence achieving predictive or criterion validity.

And if we can say, all optimal models are *equivalent* to our derived or trained model, then we can stop, as Hilbert did in his axiomatization of geometry, claiming that other axiomatizations would be equivalent in some sense.

Unfortunately these bits and the inputs and the process are something we understand by applying logic and the computer metaphor, all of which is not self-evidently applicable. Evolution has done its own thing, filtering the randomness of historical mutations with the logic of survival and natural and sexual selection. What guarantee do we have that organism-internal process reflects our logical computerological metaphorical way of understanding it? Chemical pathways are innumerable, complex, inconceiveable. So we don't really know.

On the other hand, the bitter logic of evolutionary survival, die or live, flourish or do not, is tied to the functional logic of environmental opportunity and constraint and danger: a 55mph cheetah and a 57mph springbok equals survival on average for the deer, until some cheetahs luckily get born with faster-twitch muscles or some random fitness advantage. The functional logic, whereby flourishing or not is the sufficiently-typical outcome, is the essence, even though it comes after randomness of mutation and layers of survival-filtering in and out of the womb. It is the functional logic where the world categorizes cheetahs and springboks as fast enough or too slow by putting them to each other's test and allowing physics to decide if there is flourishing or not: that logic is what makes evolved characters have their character, modulo free-rider characters.

Says Tom.