In my earlier essay, Living Fuzzy: An Incomplete Exploration of Fuzzy Logic, I began an examination of fuzzy logic and how it may be relevant to living the kind of life I want to live and being the kind of person I want to be. Well, a few months have passed and I think it’s time to return to some of those musings and investigations.
In that earlier essay, I explained that it was actually a bit of reading about my 2006 Mitsubishi Outlander’s INVECS transmission that led me to this concept of fuzzy logic. It was an entertaining and moderately insightful investigation which actually made me even more fond of the old Outlander and its transmission. But, as life brings around different priorities at different times, the time for the Outlander to leave our family arrived a short time later. On June 1, 2022, my wife and I accepted possession of a new 2022 Mitsubishi RVR.
So, in celebration of the new vehicle, here we have some further explorations of fuzzy logic!
At some point during the winter of 2021/2022, we decided that 2022 would be the year that we acquired a vehicle to replace the aging 2006 Mitsubishi Outlander. The car simply wasn’t leaving us confident of reliable transportation on the longer journeys that we wanted to be able to do. Given the probability that gasoline-powered vehicles will probably be outmoded by 2030, it may be reasonable to predict that the newly acquired RVR will be the one of the last gasoline-powered vehicles in our possession (more about that in my nostalgia-driven essays on My Next Bike and My Next (Final) Car).
While I was somewhat reluctant and disappointed to give up the Outlander, the RVR feels like a genuine and natural progression of the automotive experience. In fact, despite the convoluted nomenclature associated with marketing different products in different global jurisdictions, it is reasonable to argue that the RVR is the most recent iteration of a common design intention.
The vehicle which I purchased as the Mitsubishi Outlander was sold in other markets under the names Airtrek and was based on a design concept called ASX (Active Sports Crossover). It had a 2,625 mm wheelbase, a 2.4 litre 4G69 MIVEC engine matched to the INVECS-II fuzzy logic transmission (yes, we’re getting there).
Later in 2006 and ’07, the vehicle sold in Canada as an Outlander was a significantly larger and quite different vehicle. Meanwhile, Mitsubishi was showing another ASX concept which was more consistent with the first generation. But Mitsubishi called it RVR instead. Complicated as it may be, the RVR is actually called Outlander Sport in some parts of the globe.
It is interesting that corporations may use the same words to denote different things based on their needs; they’ll also use different words to denote the same thing. A rose by any other name is still a 4-door hatch-wagon with a 2600(ish)-wheelbase.
The 2022 RVR is essentially the second generation of the ASX concept and has a 2,670 mm wheelbase. The one we’re now driving has a 2.0 litre 4B11 MIVEC engine matched to an INVECS-III fuzzy logic continuously variable transmission.
In the earlier essay, I recommended that one learn how to operate a manual transmission. And I meant that recommendation both literally and figuratively (or metaphorically). In context of Zensylvania, automobiles are an entertaining metaphor and anchor for exploring the self. Learning to operate a transmission for a vehicle gives a person a more thorough understanding of the machines and what it means to move an object weighing hundreds or thousands of pounds down the road. An average motorcycle weight a few hundred pounds while the average vehicle weight of a car today is easily 3500 to 4000 pounds. That’s a lot of glass, metal and plastic, isn’t it? Yet many people seem to have a motivation to take as little responsibility for understanding all that technology as they can while still getting around for their business.
Similarly, responsibility for understanding how the gears of the self work is another thing that many (if not most) people seem to avoid. It’s just easier to get around on auto-pilot.
The transmissions of both of the Mitsubishi’s we’ve owned use fuzzy-logic in their electronic components. That fact reminded me of one of Robert Pirsig’s most famous observations in Zen and the Art of Motorcycle Maintenance which goes:
“The Buddha, the Godhead, resides quite as comfortably in the circuits of a digital computer or the gears of a motorcycle transmission as he does at the top of a mountain, or in the petals of a flower. To think otherwise is to demean the Buddha, which is to demean oneself. “
Well in my Mitsubishi ASX vehicles, I’ve had opportunity to observe the Buddha – or at least the concepts that aligns to the Buddha, in both the circuits of a computer and the mechanical operations of a vehicle’s transmission. Once with what I’d call a traditional gear-based transmission and now with a belt-based continuously variable transmission. These have been very different experiences.
Mitsubishi Says That
“INVECS-III CVT achieves low fuel consumption and a smooth ride
INVECS = Intelligent & Innovative Vehicle Electronic Control System
CVT = Continuously Variable Transmission
INVECS-III is an advanced system that automatically selects the optimal gear ratio based on road and driving conditions (“optimal control”), and utilizes “learning control” to match the particular driver’s driving style. In addition, a CVT that brings out the efficiency from the engine performance is provided. Like a conventional automatic transmission, there is no jolt when shifting gears and every time the accelerator is applied there is enjoyable, smooth acceleration.
Furthermore , a torque converter enables creep forward driving (slight deceleration) when the accelerator is not applied and hillstarts are made fun.
The Continuously Variable Transmission makes for optimal driving pleasure by downsizing the pulley piston, reducing the oil pump discharge rate, and controlling direct torque control. All of this results in efficient engine output, offering drivers an exceptionally smooth ride. Based on driver demand as measured by accelerator travel information, optimal efficiency is achieved between the engine and CVT according to the motive forces experienced under driving circumstances. Supple acceleration and smooth driving feel are realized in all kinds of conditions while also improving fuel economy.
I encourage you to compare this language with what Mitsubishi had to say about the INVECS II system. The key values that seem to be emphasized in this current version seem to be: fuel efficiency, smoothness of experience and enjoyment. It would have been far mor succinct for Mitsubishi to claim: INVECS III is smoother, more efficient and more fun than the INVECS II. Mitsubishi’s version however, implies all the complexity which lies in the machinations of a transmission. Of moving thousands of pounds of metal, glass and plastic (plus some occupants) down the road.
What the Hell Does all that have to do with Fuzzy Logic?
In the first part in this series, I used the explanation that fuzzy logic is a system that allows truth values that range between zero and one. With fuzzy logic, there are degrees of truth. I contrasted this with Boolean algebra where things are either zero or one. Either Completely false or completely true. Black or white. On or off. It seems to me that fuzzy logic allows for an understanding of the world that is potentially more accurate and representative than the Boolean alternative.
Particularly as it comes to coping with a world where things are far more dynamic, circumstantial and changing than a black-and-white approach can accommodate. So the fuzzy logic concept seems worthy of further exploration.
In exploring Fuzzy Logic, I necessarily need to begin at the most basic of information. I am not an expert. From the Geeks for Geeks website, I leaned that Fuzzy Logic architecture contains four parts: the rule base, the fuzzification, the inference engine and the defuzzification.
The rule base is a collection of if/then scenario propositions. In the computer programming world, these seem to be the predetermined situations that the computer is expected to encounter. In a vehicle’s transmission, perhaps these might be something like “If slippery road, 4000 RPM engine speed and aggressive driver, then final drive ratio X”. Undoubtedly, it’s far more complicated that this, but for our purposes here (which is not to become fuzzy logic programmers) let’s start there. In understanding the self, that might be “If stressful conditions, thing are happening fast and I’m angry, then behavior X”.
Given that the model allows for a variance not only in the specified variables but also in their degree or scale of intensity, it seems clear that fuzzy logic rule bases provide a more sophisticated decision-making system than a system based on absolutes as in Boolean yes/no scenarios.
Clearly also this seems to match how living organisms actually work on a biological level as well as being a more sensible approach to conscious decision-making. I can’t help but reflect on the so-called “though experiments” intended to challenge and examine a person’s moral or ethical decision-making. Morality that is based on yes/no absolutes are probably more prone to problems than morality that is sensitive to a dynamic and shifting reality with complex underlying factors. Factors that are almost infinitely variable, yet still predetermined.
The second part of fuzzy logic’s architecture is fuzzification. As I understand it, this is the part of the structure that passes raw sensor data into the rule base for assessment. In my depiction above, fuzzification takes the identified criteria (eg. road condition, engine speed, driver aggression) and applies it to the predicted (predetermined) collection of scenario propositions.
In real life, fuzzification is collecting all the variables that affect our behaviour all the way down to their basic measurable reality. Stress? How much stress as measured perhaps by certain chemicals and molecules are present in the system. It truly is marvelous that our minds and bodies already automatically do this. After millions of years of evolution, our incredibly complicated systems have a fuzzy set catalog of the if/then scenarios we are likely to encounter as a human being. That information is the measured and handled, for the most part, without our conscious decisions. For most of us its mostly a smooth, efficient and (intermittently?) enjoyable ride.
The third part of fuzzy logic architecture is the inference engine. As a slight aside, I love that term because it reminds me that language, the symbolic representation of other things (both real and imagined) contains word-meanings which are both denotative and connotative. Word meanings can be explicit or inferred.
Fuzzy logic inference engines establish the degrees of truth based on the raw data and the fuzzification process. The inference engine says which of all the possible scenario propositions are happening at any given time. The inference engine decides what is real and what is not at any given moment.
The final part of the fuzzy logic architecture is the defuzzification. In essence this it the final decision. After all is said and done, the final answer. How much truth is there and what is the appropriate behaviour or outcome?
The fuzzy logic control system is believed to provide human-like thinking which provides acceptable reasoning without necessarily being accurate, to emulate human inference and to manager uncertainty. In evaluating this, it occurs to me that the fuzzy logic system imitates human organism behaviour before thinking rather than the thinking itself. But I do think the model provides an valuable set of tools for evaluating what may be happening under the surface of ourselves and even to structure our decisions to understand what the human animal’ evolution has developed as the predictable scenario propositions, what the relevant data might be, how that data is fuzzified and then defuzzified into the actions we take.
And how we may, as a species be in a world that does not necessarily match our evolved scenario propositions. Like driving a 4000 pound construction of glass, metal and plastic down a slippery road.
Disadvantages of Fuzzy Logic Systems
For computer programmers, fuzzy logic offers some problems or disadvantages. This seems also to be true for the designers of ethical/moral thought experiments. The Geeks for Geeks article says “Many researchers proposed different ways to solve a given problem through fuzzy logic which leads to ambiguity. There is no systematic approach to solve a given problem through fuzzy logic.“
This absence of systemic approach…this ambiguity actually seems to be a strength for anyone navigating life. There’s a tension here…the whole point of fuzzy logic (hence why its termed fuzzy in the first place) is the ambiguity.
Similarly the computer programmer’s disadvantage that “Proof of its characteristics is difficult or impossible in most cases because every time we do not get a mathematical description of our approach.” is not a problem for navigating human life because human life is, for the most part, not navigated via mathematics. Indeed the fact that “fuzzy logic works on precise as well as imprecise data” is not a problem, it’s a definite benefit. Because most of life is not a matter of accuracy, it is a matter of probable survivability.
It is a matter of efficiency, smoothness and enjoyment.
A few weeks ago, an employee from Google claimed that one of it’s Artificial Intelligence systems had achieved self-awareness. Sentience. While I’m largely skeptical that this is so, it is a reminded that earth is much closer to the emergence of a new, electronics-based form of intelligence than most of us are adequately prepared to consider. Even if Google’s software is not yet there…something will emerge in the near future.
Via fuzzy logic, we can see that fuzzy control is “a technique to embody human-like thinkings into a control system.“. Clearly we humans value our deep and poorly understood brain systems so much, we want our computers to emulate them.
All the while, we have been working out ways to improve on our human use of reasoning and logic. To make use of science and reliable data. Eventually these two applications will have a meeting of their paths.
In Robert Pirsig’s second book, Lila: An Inquiry into Morals, he argues that the conceptual path that Phaedrus took eventually meets up with the path that another scholar had been taking from a different direction. That the two paths meet in the middle and complete each other. Whether this is an accurate depiction or not, I have not examined. However, it is an interesting reflection on the potential that human evolution and the development of machine learning could have a mutually-completing intersection.
The Buddha and Godhead may well, one day reside in the gears of a motorcycle transmission.
If a computer program has very little in the form of physical reality…it is encoded patterns of energy as information…yet may predictably become sentient..this is the godhead.
Computer programs are not bound by a particular physical space. They can be copied. What, then is the potential Digital Buddha. And what is there, therefore to assume that human Godhead is necessarily bound by physical human bodies. Transhumanism.
At the intersection point of humans cultivating logic and interfaces with digital devices and AIs using fuzzy logic to become sentient…there is the point when a new symbiotic Buddha may become possible.
When we are in a position to use symbols and information from digital formats as direct-access neural connections..and why aren’t our eyes exactly that type of system?…it is the storage of memories and the convergence of the simultaneously experienced digital-self and biological-self that will then possibly lead to the ability to transfer the self irrespective of the physical host.
Biosemiotics shows that we are information…bubbled up, layer after layer via fuzzy evolution to be humans. As we build fuzzy logic AIs, a kind of top down imitation of human evolution……digitical complexity to match our human evolutionary complexity.
Our evolutionary descendants may well own the stars and be independent of specific biological or mechanical hosting but capable of occupying either.
Google’s employee suggested that the corporation’s AI is a “child” – well it may be the metaphorical representation of a child of an early form of humanity.
Zensylvania: It’s a state of mind.
You Can Find and Follow the Zensylvania Podcast on Spotify.
Zensylvania Copyright © 2020-2022 by Eric Adriaans. All rights reserved.