eJournal: uffmm.org
ISSN 2567-6458, 24.March – 24.March 2021
Email: info@uffmm.org
Author: Gerd Doeben-Henisch
Email: gerd@doeben-henisch.de
CONTEXT
This text is part of a philosophy of science analysis of the case of the oksimo software (oksimo.com). A specification of the oksimo software from an engineering point of view can be found in four consecutive posts dedicated to the HMI-Analysis for this software.
CHANGE
AS described in part 1 of the philosophy of science analysis of the oksimo behavior space it is here assumed — following the ideas of von Uexküll — that every biological species SP embedded in a real environment ENV transforms this environment in its species specific internal representation ENVSP which is no 1-to-1 mapping. Furthermore we know from modern Biology and brain research that the human brain cuts its sensory perceptions P into time-slices P1, P2, … which have durations between about 50 – 700 milliseconds and which are organized as multi-modal structures for further processing. The results of this processing are different kinds of abstracted structures which represent — not in a 1-to-1 fashion — different aspects of a given situation S which in the moment of being processed and then being stored is not any longer actual, ‘not now’, but ‘gone‘, ‘past‘.
Thus if we as human actors are speaking about change then we are primarily speaking about the difference which our brain can compute comparing the actual situation S being kept in an actual time-slice P0 and those abstracted structures A(P) coming out of preceding time slices interacting in many various ways with other available abstracted structures: Diff(A(P0), A(P)) = Δint. Usually we assume automatically that the perceived internal change Δint corresponds to a change in the actual situation S leading to a follow-up situation S’ which differs with regard to the species specific perception represented in Δint as Δext = Diff(S, S’). As psychological tests can reveal this automatic (unconscious) assumption that a perceived change Δint corresponds to a real external change Δext must not be the case. There is a real difference between Δint, Δext and on account of this difference there exists the possibility that we can detect an error comparing our ideas with the real world environment. Otherwise — in the absence of an error — a congruence can be interpreted as a confirmation of our ideas.
EXPRESSIONS CAN FOLLOW REAL PROPERTIES
As described in the preceding posts about a decidable start state S and a vision V it is possible to map a perceived actual situation S in a set of expressions ES={e1, e2, …, en }. This general assumption is valid for all real states S, which results in the fact that a series of real states S1, S2, …, Sn is conceivable where every such real state Si can be associated with a set of expressions Ei which contain individual expressions ei which represent according to the presupposed meaning function φ certain aspects/ properties Pi of the corresponding real situation Si. Thus, if two consecutive real states Si, Si+1 are include perceived differences indicated by some properties then it is possible to express these differences by corresponding expressions ei as part of the whole set of expressions Ei and Ei+1. If e.g. in the successor of Si one property px expressed by ex is missing which is present in Si then the corresponding set Ei+1 should not include the expression ex. Or if the successor state Si+1 contains a property py expressed by the expression ey which is not yet given in Si then this fact too indicates a difference. Thus the differing pair (Si, Si+1) could correspond to the pair (Ei, Ei+1) with ex as part of Ei but not any more in Ei+1 and the expression ey not part of Ei but then in Ei+1.
The general schema could be described as:
Si+1 = Si -{px} + {py} (the real dimension)
Ei+1 = Ei – {ex} + {ey} (the symbolic dimension)
Between the real dimension and the symbolic dimension is the body with the brain offering all the neural processing which is necessary to enable such complex mappings. This can bne expressed by the following pragmatic recipe:
symbolicarticulation: S x body[brain] —> E
symbolicarticulation(S,body[brain]) = E
Having a body with a brain embedded in an actual (real) situation S the body (with the brain) can produce symbolic expressions corresponding to certain properties of the situation S.
DESCRIBING CHANGE
Assuming that symbolic articulation is possible and that there is some regular mapping between an actual situation S and a set of expressions E it is conceivable to describe the generation of two successive actual states S, S’ as follows:
Apply a Change Rule ξ of X
- We have a given actual situation S.
- We have a group of human actors Ahum which are using a language L.
- The group generates a decidable description of S as a set of expressions ELS following the rules of language L.
- Thus we have symbolicarticulation(S, Ahum) = ELS
- The group of human actors defines a finite set of change rules X which describe which expressions Eminus should be removed from ES and which expressions Eplus should be added to ES to get the successor state ES‘ represented in a symbolic space:
- ES‘ = ES – Eminus + Eplus . An individual change rule ξ of X has the format:
- IF COND THEN with probability π REMOVE Eminus and ADD Eplus.
- COND is a set of expressions which shall be a subset of the given set ES saying: COND ⊆ ES. If this condition is satisfied (fulfilled) then the rule can be applied following probability π.
- Thus applying a change rule ξ to a given state S means to operate on the corresponding set of expressions ES of S as follows:
- applychange: S x ES x {ξ} —> ES‘
- There can be more than only one change rule ξ as a finite set X = {ξ1, ξ2, …, ξn}. They have all to be applied in a random order. Thus we get:
- applychange: S x ES x X —> ES‘ or applychange(S,ES,X) = ES‘
Simulation
If one has a given actual state S with a finite set of change rules X we know now how to apply this finite set of change rules X to a given state description ES. But if we would enlarge the set of change rules X in a way that this set X* not only contains rules for the given actual state description ES but also for a finite number of other possible state descriptions ES* then one could repeat the application of the change rules X* several times by using the last outcome desribing ES‘ to make ES‘ to the new actual state description ES. Proceeding in this way we can generate a whole sequence of state decriptions: <ES.0, ES.1, …, ES.n> where for each pair (ES.i, ES.i+1) it holds that applychange(Si,ES.i,X) = ES.i+1
Such a repetitive application of the applychange() rule we call here a simulation: S x ES x X —> <ES.0, ES.1, …, ES.n> with the condition for each pair (ES.i, ES.i+1) that it holds that applychange(Si,ES.i,X) = ES.i+1also written as: simulation(S , ES, X) = <ES.0, ES.1, …, ES.n>.
A device which can operate a simulation is called a simulator ∑. A simulator is either a human actor or a computer with an appropriate algorithm.