киберполе

[avva]

Какая качественная трава!

Обратимся теперь к “киберполю LJ”.
Эквивалентом личности в нем является индивидуальный журнал или совокупность записей.
Психологическим окружением, соответственно, - комментарии на записи или/и записи в других журналах, имеющие отношение к записям рассматриваемого журнала.
Поведением (изменением поля) в этом случае будет появление ответных комментариев и записей в данном журнале.
Отсюда:

dx/dt=F(St)

где St - конкретная ситуация, dx - изменение в киберполе, dt - интервал времени.
Очевидно, что отдельное изменение dx целиком и полностью диктуется наличной, текущей ситуацией St и может быть выведено только из неё, но не из какой-либо прошлой или будущей ситуации. В соответствии с рассматриваемым принципом, отдельное изменение поля трансформирует его, то есть создает новую ситуацию St+n, которая, в свою очередь, определяет и новое изменение поля!
Например, в журнале имярек появляется запись - dx1/dt. На неё следует комментарий dy1/dt. Тем самым, создается ситуация S(x1y1)t или текущее возмущение киберполя. Последующие изменения dxn/dt+n и dyn/dt+n полностью определяются ею. Но они же и изменяют текущую ситуацию => S(xnyn)tn.

Во зажигают психологи.

Comments

[malaya-zemlya.livejournal.com]

I think what they want to say is that arrival of new comments depends only on the original posting, related postings in other jounals, and existing comments to these. It is simply false, of course, but math symbols make it also look stupid.

Such a shame, there's some legit research going on in mathematical psychology (like this one), but it's obscured by layers and layers of crap.

[(Anonymous)]

the link you gave concerned with
"experimental neurophysiology" (fMRI) not
"mathematical psychology".

[malaya-zemlya.livejournal.com]

Note that one of their propositions was that humans use a very specific algorithm (higher-order temporal difference) to evaluate their strategies of getting rewards and avoiding penalties.

Othre intersting works in math psych:
Mathematical modelling of marriage dynamics
Hyperbolic discounting and picoeconomics
Mathematical modelling of alcoholism (not by Venedikt Yerofeev)
And of course there's the entire body of mathematical linguistics.

[(Anonymous)]

>Note that one of their propositions was that humans use a very specific algorithm...

just print in Google:
    reinforcement learning caudate

well, everything could be "modeled" by math, but
it would be not a scientific modeling (building experimentally provable models), rather
"mathematical description" of smth ...

[malaya-zemlya.livejournal.com]

I am not sure I understoof your point correctly, but...
The algorithm itself is well-known and fairly widely used. It has known behavior, convergence properties and so forth. If humans indeed use it in our learning then
a) certain internal signals in the brain should exhibit the same patterns as the internal vaiables of the algorithm

b) behavior of humans in learning tasks can be adequately modelled and predicted by running the algorithm on a computer.

Both are very much testable predictions. The paper in question is concerned with testing point a), other people invesigate point b) as well.
IMHO experimental testing of essentially mathematical hypotheses about human behavior and its underlying mechanisms qualifies as mathematical psychology. But this is more of a terminology question.

[(Anonymous)]

:)
I do not have an access to the Nature magazine at the moment. I am not discussing the article.
My reference is pointing that there is a bulk of similar models since 80s. I know that this particular case of modeling is alright. Your point a) says it is scientific modeling. Point b) is concerned with "behavior" and looks like changing modeling back to pre-scientific "descriptive" style. The are too many explicit and implicit parameters for human behavior... IMHO, if it is behaviorist model for fishes behavior, we may call it "model", not yet for humans.
I just dislike the term "mathematical psychology". It sounds like Piaget's idea-fixe on using group theory and representation theory in psychology. The article you read rather contains neurophysiological model...

[malaya-zemlya.livejournal.com]

Well,I am not big fan of behaviorism myself but you can scientifically measure *some* parts of behavior: reaction times, choices, physiological reactions - the usual things. By analyzing those you can get an idea about internal sttructure.

Being a programmer, I like to think of it in terms of reverse engineering. If I get a new piece of software, I can get a pretty good idea about how it's written without using debugger and looking at the source code, just by looking at how it runs - it's behavior. This doesn't make me a Skinnerist, does it? :)