Category Archives: theory testing

OKSIMO.R – EVERYDAY SCENES – GO OUT FOR EAT – Part 2

eJournal: uffmm.org
ISSN 2567-6458, 18.November 2022 – 25.November 2022, 10:58h
Email: info@uffmm.org
Author: Gerd Doeben-Henisch
Email: gerd@doeben-henisch.de

Parts of this text have been translated with www.DeepL.com/Translator (free version), afterwards minimally edited.

CONTEXT

This post is part of the book project ‘oksimo.R Editor and Simulator for Theories’ and represents a continuation of Part 1.

CONTENT

In Part 1, the beginning of a simple example was presented, where an actor (here: ‘Gerd’) is sitting in his office, feels hungry, and imagines that he does not want to be hungry. In part 1, he decides to leave his office and go out to eat. Embedded in the mini-theory of this example, several concepts are explained: text types (ACTUAL description, TARGET description, CHANGE description), rule application, oksimo.R software contextualization, theory testing, inference testing, goal fulfillment testing, starting a simulation, and logical inference.

In part 2, the mini-theory will be completed. The story ends with the actor Gerd not feeling hungry anymore (at least not for the moment :-)).

Continuation of the story

An oksimo.R theory can be understood simply as a ‘story’, a kind of ‘script’, although this story has all the properties of a full empirical theory (more on theory below).

The story so far is simply told:

Starting point (Scene 1):
Gerd is sitting in his office.Gerd is hungry.
Target:
Gerd is not hungry.

Scene 2:
Gerd leaves his office.Gerd is hungry.
Target achievement so far: 0%.

The transition from Scene 1 to Scene 2 was only possible because a change rule was adopted which states that Scene 1 can be changed if the condition ‘Gerd is hungry’ holds. Since this is the case, the property ‘Gerd is sitting in his office’ was removed and the new property ‘Gerd is leaving his office’ was added.

For another continuation, a rule is missing at the moment. However, the only change rule so far can be reapplied over and over again, so that scene 2 is repeated any number of times (like a record player hitting a broken groove in the record, so that the record player repeats that track endlessly until we turn it off.)

This ‘repeatability’ can become a problem if you’re not careful. Here’s an example of unwanted repetition (which we ultimately don’t want!).

Unwanted repetition(s)

Since there is a Greek bistro ‘around the corner to the left’ where Gerd could eat a snack, we write down the following new change rule:

CHANGE Description 2:

IF:

Gerd is hungry.

THEN:

Add as a property to the ACTUAL situation: Gerd decides to go to the Greek around the corner.

Remove as a property from the ACTUAL situation: – Nothing -.

APPLICATION of the change description:

Since the condition ‘Gerd is hungry.’ is met, the rule could be applied and we would get the following result with this rule:

THEN:

NEW ACTUAL situation (with rule 2):

Gerd decides to go to the Greek around the corner. Gerd is hungry.

However, there is still rule 1, which does not disappear (as an option, however, conceivable). This rule has the same condition as rule 2 and can therefore also be applied. It would produce the following result:

NEW ACTUAL situation (with rule 1):

Gerd leaves his office. Gerd is hungry.

A ‘union’ of the continuation according to rule 1 and the continuation according to rule 2 leads to the following result:

Gerd decides to go to the Greek around the corner. Gerd is hungry. Gerd leaves his office.

With the oksimo.R software (level 2) this would look like this:

Entering a Change Rule

Rule:Food1-Location1
Conditions:
Gerd is hungry.
Positive Effects:
Gerd decides to go to the Greek around the corner.

Negative Effects: — Nothing —

Starting a New Simulation

With code number one you can start a new simulation. We need the following ‘ingredients’:

Selected visions:
Food1-v1
Selected states:
Food1
Selected rules:
Food1-Location1
Food1-Decision1

Protocol of the simulation (simple version)

Your vision:
Gerd is not hungry.

Initial states: 
Gerd is hungry.,Gerd is sitting in his office.

Round 1

Current states: Gerd is hungry.,Gerd leaves his office.,Gerd decides to go to the Greek around the corner.
Current visions: Gerd is not hungry.

0.00 percent of your vision was achieved by reaching the following states:
None

Round 2

Current states: Gerd is hungry.,Gerd leaves his office.,Gerd decides to go to the Greek around the corner.
Current visions: Gerd is not hungry.

0.00 percent of your vision was achieved by reaching the following states:
None

Already after two simulation cycles one recognizes that everything repeats itself. And with knowledge of the change rules one knows that both are ‘activated’ again and again as long as their condition is fulfilled. In the concrete example this is the case. This points to a general structure of rule-driven changes with situational reference.

On the meta-logic of situational change rules

At this point it should be remembered again that an ACTUAL description is nothing more than a ‘set of linguistic expressions’ of the respective language chosen. Here the English language is used. In the original source of the oksimo.org blog the German language is used. Any other language is also possible.

However, from the point of view of the respective actor working with such IS-descriptions, every linguistic expression used in the space of his ‘linguistic understanding’ has additionally a ‘special meaning’, which can partially be ‘correlate’ with ‘properties of the external body world’ in a ‘specific way’. So, if someone reads the expression ‘Gerd’, he will mostly associate with it the idea that it is the ‘name of an individual’. And when one reads the linguistic expression ‘… sitting in his office’, one will usually think of a ‘room in a building’. Both notions ‘name of an individual’ as well as ‘room’ in a building’ have – normally – the property that one can ‘relate’ to them concrete ‘objects of the external body world’ via ‘individual perception’. This can happen in many ways, e.g. in which someone else says to me “Look (and he points to a person), this is Gerd”, or I come into the room 204 in building 1 of the Frankfurt University of Applied Sciences and someone says to me “Look, this is Gerd’s office”. In both cases, a concrete perception can then connect with an ‘imagined conception’ in such a way that the inherently ‘abstract’ conception of an individual person in a room connects (associates) with a bundle of sensually perceived properties.

With this background knowledge one can then understand why an IS-description as a set of linguistic expressions has ‘two faces’: (i) At first sight there are only a set of linguistic expressions without any recognizable further property, and (ii) , starting from the linguistic expressions, mediated by the linguistic meaning knowledge of a speaker-hearer of the respective language, a set of meanings appears, which in the case of an IS-description must by agreement all have at least one concrete reference to the external body world. Roughly, one can therefore say at this point that every linguistic expression of a normal language can be linked (associated) with a ‘property’ of the external body world. In this second sense, an ACTUAL description then represents not only a ‘set of linguistic expressions’ but at the same time also (language comprehension in the actor presupposed) a ‘set of body-world properties’. The removal of a linguistic expression then means at the same time the removal of a property, and the addition of a linguistic expression the addition of a property.

Due to this generally assumed ‘linguistic dimension of meaning’ in each involved actor, ACTUAL descriptions thus potentially represent a connection between the virtual images in the brain of an actor to possible sensually perceptible correlates of an external body world linked to it, for which a ‘self-driven dynamic’ is assumed. By this is meant that the world of our sensual perception (linked with our memory!), apparently constantly ‘partially changes’ and simultaneous ‘partial stays constant’. The ‘extension’ of the ‘quantity of the properties of the external body world’ seems to be almost ‘infinite’ and at the same time also the possible extent of the changes.

Against this background (largely always hypothetical), any ACTUAL description always appears as a ‘very small selection’ of this body world property set and a concrete ACTUAL description forms a kind of ‘snapshot’ of a continuously dynamic event which can only be ‘traced’ in a highly simplified way via the explicitly formulated rules of change. In particular, there is a problem of how to keep an ACTUAL description ‘up to date’ when the external body world is continuously changing due to its ‘inherent dynamics’ without any oksimo.R theory-builder actor having formulated a single rule of change. In other words, an ACTUAL description ‘becomes obsolete’ by itself if the ‘coupling’ of the ACTUAL description to the external body world is not ensured with ‘appropriate’ change rules. In order to be able to do this, one needs a ‘translator’ who continuously ‘maps’ the changes of the external body world into the linguistic meaning space of the actors and these then generate corresponding linguistic expression sets.

Further possible requirements for a process

After these meta-logical considerations about the function of ACTUAL descriptions in the interplay with an assumed external body world with its own inherent dynamics, some further aspects shall be brought up here, which are/can be significant for the creation of a ‘plan’.

So far the small oksimo.R theory – the current story – has the following format:

Initial state (Scene 0): 
Gerd is hungry.Gerd is sitting in his office.

The vision:
Gerd is not hungry.

Scene 1:
Gerd is hungry.Gerd leaves his office.Gerd decides to go to the Greek around the corner.

Success: 0.00 percent 

Scene 2:
Gerd is hungry.Gerd leaves his office. Gerd decides to go to the Greek around the corner.

Success: 0.00 percent

The goal is still that the actor Gerd reaches his goal, the ‘Greek around the corner’, so that he can eat, for example, so that his feeling of hunger disappears.

For this, on the one hand, there must be rules that move the actor ‘through space’ to the ‘Greek around the corner’, on the other hand, the rules must be such that they cannot activate properties that should no longer occur in the process at all.

A rule like ‘Food-Location1′, which ensures that Gerd leaves his office, should not be applied again at a ‘later time’, similarly the rule ‘Food1-Decision1’, which describes the decision that Gerd wants to go to the ‘Greek around the corner’.

Since the activation of a change rule depends on the respective ‘condition’, this means that the condition for a rule should be such that the ‘triggering property’ is as ‘process-specific’ as possible. For the property ‘Gerd is hungry’, which is valid throughout the whole story until the actual eating, this is rather not true. Since all rules with this ‘non-specific trigger’ would be activated again and again, until at some point the eating produces the new property ‘Gerd is not hungry’.

This raises the question of how an ACTUAL description should be formatted such that, in addition to ‘long-living’ properties, there are also ‘short-living’ properties that can actually serve selectively as ‘triggers for rule activation’.

Time information is often not enough

In everyday life we are used to link events to a certain time, thereby assuming the existence of clocks that are synchronized worldwide; or the whole thing extended by a calendar with days, weeks, months and years. Such a tool can easily be introduced into an oksimo.R theory. But this solves the problem only partially. For many events one knows in advance neither ‘whether’ they occur at all, nor ‘when’ this will happen. In that case, the only possibility is to link a ‘subsequent event’ directly to a certain ‘preceding’ event: For example, it only makes sense to open the umbrella when it actually rains. There is usually no exact date when this event will occur.

Design perspectives: Goal and precision

What use are these considerations in the specific example where a ‘sequence’ is sought that leads to Gerd experiencing that his feeling of hunger disappears?

Two general considerations may be helpful here:

  1. Thinking from the end (goal)
  2. What ‘accuracy’ is required/desired?

If one knows a goal (which is not self-evident; often one first has to find out what a meaningful goal could be), then one can try to think ‘backwards’ from the goal by being guided by the question, ‘Which action A do I have to do to achieve result B?’. In the case of the desired goal state ‘Gerd is not hungry’, the usual experience would be to eat something ‘appropriate’, which leads to the ‘disappearance of the feeling of hunger’ (most of the time). Then you have to know what that ‘food’ might be, where to get it, and what you would have to do to get there (let’s ignore the case of someone just bringing something from home to eat). From such ‘backward-thinking’ a hypothetical sequence of actions can emerge, which can become the basis for a ‘plan’, which the actor will work out ‘in his head’ and then implement piecemeal by corresponding ‘real actions’.

The question of ‘accuracy of representation’ (of a story, of a theory) is not easy to answer. If engineers have to program a robot that is supposed to be able to perform certain operations, then this will normally require an almost merciless accuracy (apart from the case that there are already many ready-made modules that can take care of ‘small stuff’ (such as so-called ‘machine learning’ after successful training)). If it is the author of a crime novel or the author of a screenplay, then besides ‘factual aspects’ very much also the ‘effect on the readers / viewers’ must be considered. In the case of achieving a concrete goal in a concrete world, the potential success of the implementation of a description depends entirely on whether the concrete requirements of the world – here the everyday world – are completely satisfied. Of course, the reader/listener/user of a description also plays a major role: If we can assume that we are dealing with ‘experts’ who ‘know’ the process to be performed well, we can perhaps work with hints only; if we are dealing more with ‘newcomers’, then we must provide very detailed information. Sometimes a purely text-based description is not sufficient; more is then needed: pictures, videos or even your own training.

With a target and with ‘everyday’ accuracy

In the concrete case, there exists a target and ‘everyday experience’ is to be taken as a yardstick for accuracy; the latter, of course, leaves much ‘room for interpretation’.

Starting from the goal ‘thought backwards’ the following chain of actions seems plausible as a ‘hypothetical plan’:

  1. Gerd is not hungry’ because:
  2. ‘Gerd is eating his stew’ because:
  3. ‘Gerd gets his order’ because:
  4. ‘Gerd is ordering a stew’ because:
  5. ‘Gerd is standing in front of the counter’ because:
  6. ‘Gerd enters the bistro’ because:
  7. ‘Gerd goes to the Greek around the corner’ because:
  8. ‘Gerd decides to go to the Greek around the corner’ because:
  9. ‘Gerd is hungry’, ‘Gerd is in his office’, because:
  10. … there is a ‘cut’ here: arbitrary decision where to start the story/theory …

In fact, at any moment, there is not only one choice, and many things can happen during the ‘execution’ of this ‘plan’, which can result in a change of the plan. And, of course, there are many more possible aspects that could (or should) be relevant for the execution of this plan.

Constant and variable properties

As observed earlier, there are properties that are ‘rather constant’ and those that are ‘short-lived’. For example, in the context of the ‘plan’ above, the property ‘Gerd is hungry’ is constant from the beginning until the event ‘Gerd is not hungry’. Another property like ‘Gerd leaves his office’ is rather short-lived.

If we take the above hypothetical plan as a reference point, the following distribution of ‘rather constant’ and ‘rather short-lived’ properties suggests itself (left column ‘rather constant’, right column ‘rather short-lived’):

Gerd is hungry.Gerd is in his office
Gerd is hungry.Gerd decides …
Gerd is hungry.Gerd walks …
Gerd is hungry.Gerd enters …
Gerd is hungry.Gerd stands in front of ..
Gerd is hungry.Gerd orders …
Gerd is hungry.Gerd gets …
Gerd is hungry.Gerd eats …
Gerd is not hungry.

A simple strategy to avoid inappropriate repetitions would be the one in which the condition of a change rule refers to a ‘rather short-lived’ property that ‘automatically’ disappears with the implementation of a change rule.

Example (short form):

  1. If: ‘Gerd is hungry’ and ‘Gerd is in his office’, Then: ‘Gerd decides to…’.
  2. If ‘Gerd is hungry’ and ‘Gerd decides…’, Then add: ‘Gerd goes…’, Delete: ‘Gerd in office…’
  3. If ‘Gerd is hungry’ and ‘Gerd goes…’, then add: ‘Gerd enters…’, delete: ‘Gerd goes…’
  4. If ‘Gerd is hungry’ and ‘Gerd enters…’, then add: ‘Gerd stands in front of…’, delete: ‘Gerd enters…’.
  5. If ‘Gerd is hungry’ and ‘Gerd stands in front of…’, then add: ‘Gerd orders …’, delete: ‘Gerd stands in front of …’
  6. If ‘Gerd is hungry’ and ‘Gerd orders …’, then add: ‘Gerd gets …’, delete: ‘Gerd orders …’
  7. If ‘Gerd is hungry’ and ‘Gerd gets …’, then add: ‘Gerd eats…’, delete: ‘Gerd gets’.
  8. If ‘Gerd is hungry’ and ‘Gerd eats’, then add: ‘Gerd is not hungry’, delete: ‘Gerd eats…’.

This small example already shows very clearly the ‘double nature’ of our everyday reality: one is what we do ourselves, and the other is the ‘effects’ of our doing in the external body world. When someone intends to ‘walk’ and then actually walks, then one moves the body, which ‘automatically’ changes the position of the body in the external body world. Normally, one does not describe these ‘effects’ explicitly, because every person knows that this is so, based on everyday world experience. But if one wants to create a ‘description’ of the external body world with its properties, which is such that an ACTUAL description contains everything that is important for the description of a process, then one must also make some of the ‘implicit properties’ ‘explicit’ by including them in the description. Most important is the attention to ‘more ephemeral’ (temporary) properties, whose presence or absence is crucial for many actions.

Simulation extension

The extended simulation adopts the action outline from ‘backward thinking’ (see above). New change rules are formulated for this purpose.

The previous ACTUAL description is retained:

Eat1

Gerd is sitting in his office.
Gerd is hungry.

The current TARGET description is retained:

Eat1-v1

Gerd is not hungry.

The following change rules are reformulated:

Eat1-Decision1

Rule name: Eat1-Decision1
Conditions:
Gerd is hungry.
Gerd is sitting in his office.
Effects plus:
Gerd goes to the Greek.
Gerd decides to go to the Greek restaurant around the corner.
Effects minus:
Gerd is sitting in his office.

Eat1-Enter1

Rule: Eat1-Enter1
Conditions:
Gerd goes to the Greek.

Positive Effects:
Gerd is in the bistro.
Gerd enters the bistro.

Negative Effects:
Gerd goes to the Greek.

Gerd decides to go to the Greek restaurant around the corner.

Eat1-Stand-Before1

Rule: Eat1-Stand-Before1
Conditions:
Gerd enters the bistro.
Positive Effects:
Gerd stands in front of the counter.

Negative Effects:
Gerd enters the bistro.

Eat1-Order1

Rule: Eat1-Order1
Conditions:
Gerd stands in front of the counter.
Positive Effects:
Gerd orders a stew.

Negative Effects:
Gerd stands in front of the counter.

Eat1-Come1

Rule name: Eat1-Come1
Conditions:
Gerd orders a stew.
Effects plus:
Gerd gets his stew.
Effects minus:
Gerd orders a stew.

Eat1-Food1

Rule: Eat1-Food1
Conditions:
Gerd gets his stew.
Positive Effects:
Gerd eats his stew.

Negative Effects:
Gerd gets his stew.

Eat1-Not-Hungry1

Rule:Eat1-Not-Hungry1
Conditions:
Gerd eats his stew.
Positive Effects:
Gerd is not hungry.

Negative Effects:
Gerd is hungry.
Gerd eats his stew.

Collecting single Rules in one Rules Document

If you wanted to start a new simulation now, you would normally have to enter each rule individually. When experimenting, this can quickly become very annoying. Instead, you can combine all rules that ‘thematically’ ‘belong together’ in a ‘rule document’. Then you only need to enter the name of the rule document in the future.

In the present case, a rule document with the name ‘Eat1-RQuantity1′ is created. This document then includes the following rules:

  1. Eat1-Decision1
  2. Eat1-Enter1
  3. Eat1-Stand-Before1
  4. Eat1-Order1
  5. Eat1-Come1
  6. Eat1-Food1
  7. Eat1-Not-Hungry1

To start a new simulation, you then only need to enter the following:

Selected visions:
Eat1-v1
Selected states:
Eat1
Selected rules:
doc Eat1-RQuantity1

Enter maximum number of simulation rounds

>10

SIMULATION PROTOCOL (with rule applications)

Simulation saved as: Eat1-sim6

Your vision:
Gerd is not hungry.

Initial states: 
Gerd is hungry.,Gerd is sitting in his office.
Initial math states

Round 1

Current states: Gerd is hungry.,Gerd decides to go to the Greek restaurant around the corner.,Gerd goes to the Greek.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 2

Current states: Gerd is hungry.,Gerd is in the bistro.,Gerd enters the bistro.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 3

Current states: Gerd is hungry.,Gerd stands in front of the counter.,Gerd is in the bistro.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 4

Current states: Gerd is hungry.,Gerd orders a stew.,Gerd is in the bistro.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 5

Current states: Gerd is hungry.,Gerd gets his stew.,Gerd is in the bistro.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 6

Current states: Gerd is hungry.,Gerd is in the bistro.,Gerd eats his stew.
Current visions: Gerd is not hungry.
Current values:

0.00 percent of your vision was achieved by reaching the following states:
None

Round 7

Current states: Gerd is not hungry.,Gerd is in the bistro.
Current visions: Gerd is not hungry.
Current values:

100.00 percent of your vision was achieved by reaching the following states:
Gerd is not hungry.,

Further Reading:

We recommend to continue with the explanation box about ‘World, Space, Time’ and then, after having read this, go to ‘Daily Routine (temporal structures)’.