Philosophy Essay: Against Representation

**This was for my ARTI/PHIL/PSYC 3550 class as my final essay and in summary, this essay argues against the classical models of cognitive science where representation is seen as essential to artificial intelligence. Final grade was a B+, so there are points where there might be thin spots, but overall, if y’all have any thoughts or points of disagreements, leave’em in the comments below. :)**

Introduction

Throughout much of our study and development of human cognition and its replication in the various forms of artificial intelligence, there has been an underlying assumption from which we have based our work on– that to replicate intelligent thought and intelligent behavior requires extensive representation. However, representation isn’t only unnecessary but that it would actually be detrimental to our efforts to create true artificial intelligence and our understanding of our own cognition if we keep the level of representation that we currently implement in our machines.

Embodied Cognition

“A machine can exist in an abstract plane where it can crunch numbers detached from the physical world. We, on the other hand, are enmeshed in it.”

A key difference between us and the machines we create is in our embodiment. A machine can exist in an abstract plane where it can crunch numbers detached from the physical world. We, on the other hand, are enmeshed in it. Our bodies constantly feed us information about the environment and we have no way of disconnecting from it. Our cognition is fundamentally embodied and our bodies affect the way we think and vice versa. 

This can be shown in our use of language. Language is a cognitive tool that both permits the expression of and the limitations of our cognition and how we use it tells us a lot about how we think. Since so much of our cognition is expressed and shaped within the constraints of our language, Language shows us just how much of our cognition is rooted in our bodies. For example, warmth is associated with affection (“warming up to someone”), weightiness is associated with value and importance, and exposure to immoral or unethical instances causes a feeling of uncleanliness [1]. In these cases, it shows that there is a reciprocal relationship between our cognition and our physiological form. 

This strong connection between body and mind means we often never have to completely hold our thought processes in our brains. For example, if we are working out a math problem, we can write it down and refer back to earlier steps to help complete later steps. This way, we only have to store the current step of the process in our brain, make the immediately relevant calculation and then the information can be stored and kept track of on an external medium. Since our brains are limited in its energy stores, it saves a lot of energy by processing information in small chunks and to externalise it like this. Rather than wasting time and energy replicating a model of the problem space internally to manipulate, we can just reference reality to inform us on what to do next. This ties back to cognitive technology [2] where external mediums can be used to bolster our cognitive processes and therefore, become an extension of our cognition. It makes sense, then, that since we can save energy, increase our cognitive abilities and make use of a body we were born with, a lower level of representation and a more embodied cognitive process will allow us to make better use of the body we possess and our energy resources and therefore, is a much more efficient way than the fully representational classical models. We will see how this need for efficiency means that, in our brains, just the minimal level of representation is used to allow us to “scrape by”.

Efficiency Argument

The human brain and its mysterious inner workings that we try to replicate in artificial intelligence have been shown to use far less representation than we thought. We can see this in all the ways that the brain fails to notice what should have been significant details like in the famous Gorilla Experiment [3] where participants, upon being asked to watch a clip of a basketball game, fail to spot when a person in a gorilla suit walks through the middle of the game. If the brain used the same level of representation as our machines, then all the participants should have made a complete internal model of the basketball game and noticed the gorilla. Instead, it seems that the brain is selective in its attention and by doing so, restricts the amount of information it needs to process at one time and is, therefore, more efficient.

” Rather, it seems that, for the most part, our brains can get by with as little representation as possible to be functional and be generally correct when it comes to problem-solving.”

Besides incomplete representation, the brain is also prone to a phenomenon called gist memory[4] where our memory can be, at times, approximative and at worst, unreliable. In a task where participants are asked to remember words with similar associations like “ice”, “snow”, and “winter” from a list, participants often say they remember a word, like “cold”, that wasn’t present on the list but also shared those associations. Other shortcomings like the notoriously unreliable eyewitness testimony have further exposed just how little our brain represents from the world. Rather, it seems that, for the most part, our brains can get by with as little representation as possible to be functional and be generally correct when it comes to problem-solving.

Evolution Argument

However, we are not the only ones capable of exhibiting intelligent behavior. In fact, much of intelligent behavior doesn’t need a brain but rather an interlocking system of simple operations that, when viewed gestalt, suggest intelligence. Roboticist Rodney Brooks coined the term “subsumption architecture”[5] to describe such a structure. We need to look no further than our own bodies for an example. Our immune system possesses remarkable adaptability and complexity[6] that, if we didn’t know any better, we might have thought that the individual cells within our body possess their own intelligence. From the outside, the operations of the cells that make up our immune system seem to be a conscious, coordinated effort but each cell is actually just responding to certain stimuli released by other cells like proteins or hormones. 

This sort of organisation can be seen in organisms like ants where they use very simple chemical markers that elicit simple behaviors that, when repeated within a colony of thousands or millions of members, can forage for food, wage war, maintain fungus farms and aphid herds, and build floating rafts to survive floods. Brooks himself built creatures based on crickets that showed how intelligent behavior like finding mates which require pathfinding and spatial reasoning can actually be the result of very simple physiochemical reactions within the body[5].

Therefore, not only is representation not needed at all for certain intelligent behaviors but even for the human brain, representation is only very minimally used. However, there is more to our intelligence than merely problem-solving and so far, it doesn’t seem like they’ve been accounted for. 

Rebuttals

Imagination, for one, hasn’t been accounted for. Imagination, according to Merriam-Webster, is “the act or power of forming a mental image of something not present to the senses or never before wholly perceived in reality”. This means that, necessarily, to be able to imagine means that the things we imagine must not already exist and therefore unable to be represented. It can be argued that our imagination is actually just a composite of previous experiences (representations) put together in novel ways to make something technically “new”. However, it must be considered that just because representations of something exists doesn’t mean that it’s necessarily real. For example, illusions and hallucinations causes a person to represent something that isn’t actually there. I am of the thought that the false representation of a non-actual object cannot be truly considered to be represented the same way we represent what’s real[see 7]. In the same vein, although I cannot yet explain the presence of these false representations, I argue that imagination, due to its focus on the non-actual, does not in fact rely on representations or at least, the type of representation as we currently understand it to be.

“We don’t know why it is so, it just feels right.”

Second, our memory, it might seem, also depends on representations. After all, memory is formed through the encoding of past experiences. However, I argue for limited representation, not the total elimination of representation. It is true that a certain type of memory (fact-based, non-phenomenal memory) may be represented but it still doesn’t account for other types. Muscle memory, for example, is formed through repetition of a movement and once formed, is accessed without conscious effort. This is different from the explicit content that makes up represented memories. Similarly, ability-based memory also seems to lack representation. For example, when teaching another to drive, it is hard to articulate the feeling of the car, how to know how far to turn or how far the hood extends past the steering wheel. Perhaps a better example can be found in our use of language. We can usually go about our day and communicate with no problem but once we are forced to slow down and explain the finer details of grammar or convention, we are often stumped. We don’t know why it is so, it just feels right. Efficiency once again plays a role here. By cutting out the thinking part of certain operations like driving a car or figuring out grammar before speaking, the brain doesn’t need to waste resources “reinventing the wheel” and rather, just knows that certain stimuli should entail certain reactions. You don’t need to think, “the light is green and green means go” before pressing the gas pedal. You do it automatically. This way, the process cuts out thinking and representing entirely and can go straight from sensing to reacting.

It was given for a long time that the representational theory of mind would be the basis on which we can produce higher cognition in our own creations but now we know better than to let that be the end-all-be-all. With a better understanding of the reciprocal relationship between our bodies and our cognition, studies that reveal just how little our own brains rely on complete representation, and the fact that not everything that looks intelligent is intelligent, it seems that representation is actually a rather inefficient and insufficient explanation for all the miracles we are capable of.

References

[1] McNerney, Samuel. “A Brief Guide to Embodied Cognition: Why You Are Not Your Brain.” Scientific American Blog Network, Scientific American, 4 Nov. 2011, blogs.scientificamerican.com/guest-blog/a-brief-guide-to-embodied-cognition-why-you-are-not-your-brain/.

[2] “Chapter 9: Extended Minds?” Mindware: an Introduction to the Philosophy of Cognitive Science, by Andy Clark, Oxford University Press, 2014, pp. 192–211.

[3] Simons, Daniel, director. Selective Attention Test. YouTube, YouTube, 10 Mar. 2010, www.youtube.com/watch?v=vJG698U2Mvo.

[4 ]Makin, Simon. “What Happens in the Brain When We Misremember.” Scientific American, Scientific American, 9 Sept. 2016, www.scientificamerican.com/article/what-happens-in-the-brain-when-we-misremember/

[5] Brooks, Rodney A. “Intelligence Without Reason.” The Artificial Life Route to Artificial Intelligence, 2018, pp. 25–81., doi:10.4324/9781351001885-2. 

[6]Chaplin, David D. “Overview of the Immune Response.” The Journal of Allergy and Clinical Immunology, U.S. National Library of Medicine, Feb. 2010, www.ncbi.nlm.nih.gov/pmc/articles/PMC2923430/

[7] Loar, Brian. “Transparent Experience and the Availability of Qualia.” Consciousness and Meaning, 2017, pp. 273–290., doi:10.1093/acprof:oso/9780199673353.003.0016.

Symbolic Logic: Conditional/Indirect Proofs and Proving Theorems

Hey guys! So, this time, we’re going look at other methods we can use to construct proofs when just deriving from the premises isn’t enough.

Conditional Proof (CP)

The setup:

Basically, you use this method when the conclusion or a part of the conclusion you want is a conditional. This makes it so you assume the predicate in order to derive the consequent. Here’s an example:

Indirect Proof (IP)

The setup:

For this method, you use this primarily when the conclusion is a negated statement. You assume the un-negated form of the conclusion and attempt to find a contradiction so that the assumption is false, thus ending at the negated form. It also works the other way around where the conclusion isn’t negated so you make the assumption negated instead and then use the DN rule at the end. It’s also super useful when proving theorems where you have a limited plan of action. An example:

Theorems

Theorems are formulas that can be proven true without premises so the proofs for theorems have the additional challenge of not being able to build off of premises. Therefore, the above two methods are essential to be able to do proofs of theorems. Here’s an example:

RULES

  1. All assumptions must be discharged(closed).
  2. Lines between different assumptions must not cross.
  3. Once discharged, steps within the subproof cannot be used anymore.

On to the next page for a few practice problems!

Symbolic Logic: 10 Replacement Rules

This is part two of our little intro to Symbolic Logic. We’re going to expand our repertoire of rules we can employ in our proofs. These rules are all about putting logic statements into an alternative form. A lot of these rules will be familiar as they’re used in mathematics. One of the differences between these rules and the basic 8 is that these are reversible, hence the :: symbol to denote a two-way operation. The format of this post is going to be similar to the last one; the rules will be listed first, then some simple examples and then a couple of practice problems on the following pages.

1. Double Negation (D.N)

p :: ~~p

2. Commutation (Comm.)

p V q :: q V p

p • q :: q • p

3. Association (Assoc.)

[(p V (q V r)] :: [(p V q) V r)]

*applies to AND operators the same way*

4. Duplication (Dup.)

p :: p V p

5. DeMorgan’s Law (DeM.)

~(p V q) :: ~p • ~q

This law describes how a negation gets distributed into a parenthesised statement. It negates the two variables and switches the operator from an AND to an OR or vice versa. It only works on AND and OR operators though so if you have a (BI)CONDITIONAL operator inside the parenthesis, you’ll need to use one of the later replacement rules to make them into one.

6. Biconditional Exchange (B.E.)

(p ≡ q) :: [(p ⊃ q) • (q ⊃ p)]

7. Contraposition (Contra.)

(p ⊃ q) :: (~q ⊃ ~p)

8. Conditional Exchange (C.E.)

p ⊃ q :: (~p V q)

9. Exportation (Exp.)

[(p • q) ⊃ r] :: [(p ⊃ (q ⊃ r)]

10. Distribution (Dist.)

[p • (q V r)] :: (p • q) V (p • r)

[p V (q • r)] :: (p V q) • (p V r)


Examples




Alright y’all, on to the practice problems.

Symbolic Logic: 8 Basic Inference Rules

Hi y’all! So, if you’re computer science majors/philosophy major/etc., you probably have to take this class in college. I love this stuff because it’s very procedural and the proofs they give for you to solve are like puzzles and puzzles are super fun. Today, we’re gonna look at the 8 basic rules and then we’ll look at the replacement rules and more. I’m going to assume that y’all know the basic structure of sentential logic including operators and truth tables. Let’s get started.

Structure

A proof is a procedure which is supposed to derive the desired conclusion from a set of premises. To do this, the proof has to be set up in a certain way. First, all lines of a proof must be numbered. The premises make up the first lines of the proof along with the desired conclusion. Then, all subsequent derivations from the premises are listed below with the justification for each step listed along the right side, noting which rule was used and what lines of the proof were referenced. Here’s an example:

  1. Premise Pr. (for premise)
  2. Premise Pr. /:. Conclusion
  3. Derivation (Name of Rule) 1, 2 (lines 1 & 2 referenced)

For our purposes on this page, the visualisations for each of the rules below will not be written in this vertical fashion as they are cumbersome to format in the WordPress editor so it’ll be horizontal.

Let’s get into the rules and then work on some examples which will be on page 2.

1. Simplification (Simp.)

p • q /:. p

OR

p • q /:. q

If there is a conjunction, then both conjuncts can be individually represented as being true by themselves.

2. Conjunction (Conj.)

p; q /:. p • q

If two variables are true, then they can be joined in a conjunction.

3. Addition (Add.)

p /:. p V q

This rule is incredibly powerful as it allows you to introduce new elements into a disjunction as long as we have one of its disjuncts as true.

4. Disjunctive Syllogism (D.S.)

p V q; ~q /:.p

OR

p V q; ~p /:. q

If one of the disjuncts is stated to be false, then the remaining disjunct is true.

5. Modus Tollens (M.T)

p q; ~q /:. ~p

If the consequent of a conditional is false, then its antecedent is also false.

6. Modus Ponens (M.P.)

p q; p /:. q

If the antecedent of a conditional is true, then the consequent is also true.

7. Hypothetical Syllogism (H.S.)

p q; q r /:. p r

If the antecedent of a conditional leads to an antecedent of another conditional, then you can infer that the first antecedent leads to the consequent of the second conditional.

8. Rule of Dilemma/ Constructive Dilemma (C.D)

p V q; p s; q r /:. s V r

If either of the antecedents of two conditionals is true, then either of their consequents must also be true.


Alright, off to examples! We’re gonna start off easy and build onto harder and longer proofs.





Alright, now on to real problems; I have included two. I’ll give you guys the premises and the desired conclusion and I’ll post answers on how to derive the conclusion on the page after that. On to the next page!

Recent Introspections | Short Essays

Hello, everyone! Today’s post isn’t about the greater world, it will be about me. These two short essays were written for my UGA application, which was the last application I finished. I suppose you can use these as example essay for UGA’s writing portion of the application. So, this will be a bit of something to help understand me better– a memoir, if you will.

Essay #1:

The college admissions process can create anxiety. In an attempt to make it less stressful, please tell us an interesting or amusing story about yourself that you have not already shared in your application. Respond in 200-350 words.

I’ve once had a “near-death” experience involving some geese and an old, gnarly pine tree.

A mostly clear sky, a little breeze, the temperature at a balmy seventy-five degrees — a perfect spring day for climbing. Up through the crooked branches, one hand and foot at a time, I would often find my quiet place nestled among fragrant needles. There, the drudgeries of a too-ordinary life for a too-imaginative mind would fall away and I would be free to fancy myself as a character in whatever world I was currently immersed in. Only the slight stickiness of the tree’s numerous small wounds anchored me against the solid bark. It was a portal. Being neighbors, so to speak, I also had quite a few run-ins with geese around the area.

The geese were plentiful during the spring and they bring with them natural alarm clocks and fertiliser for everyone’s lawns. They also come to lay eggs and when a goose has little goslings, you better stay away or else a pecking is going to be the least of your problems.

So, on that day, I was up in the tree and it started raining. Just as I was about to hop down from the last branch, two parents and their gaggle of waddling children passed right underneath and decided to take shelter there. They noticed me but still didn’t move. It was a shock to me, I suppose, being stuck on that last branch as it began to rain harder and not knowing what to do. I don’t remember how I got down. I suppose I just jumped down and booked it before an angry mama geese had the chance to charge at me. Before I had the courage to jump, though, I stayed frozen on that last branch for what seemed like an eternity, torn between fear and urgency.

It is ironic now that I look back at it. Maybe it was a warning that whatever story I placed myself in, it isn’t the real world and that if I went too far, I might not be able to come back.

350 words

 

Essay #2:

For this essay, you had several prompts to choose from. Since I’m an artist and a writer, I chose to go with the creativity one.

Creativity is found in many forms including artistic avenues, intellectual pursuits, social interactions, innovative solutions, et cetera. Tell us how you express your creativity.

I am an artist and a writer. My chosen medium is a pencil. It only comes in one color but with a skillful hand, you can create many shades with it, enough to illustrate an entire world. With a pencil, just black squiggles, you can also write entire histories. When I’m bored, I created meaning. When I was bored, I created stories. When I was bored, I drew and wrote. This was what kept me going as a child when I didn’t know how to speak the language and was limited to staying at home and going to school and this has kept me going when I feel like I’m about to burst and not having somewhere to release.

In creating, whether it be characters or interpretative art pieces, I can give a little of myself to the real world and felt it as an affirmation of my existence because it was a manifestation of my thoughts. I think therefore I am. Books told me that I wasn’t alone in the thoughts I was having and that despite not knowing how to communicate efficiently, there were others that wrote like me and thought like me. Others that wished to create a world and develop meaning. I suppose what nurturing I missed from my family as a child was instead provided to me through the indirect sympathies of the characters that I read about. Creativity to me is something that drives me on because then I know I can affect the world permanently when I feel invisible.

When I write or draw, I try to create layers. Beyond syntax and grammar, there is subtext. Beyond lines and shadowing, there is interpretation. I try to tackle universal themes. I write about politics and history through the lens of a psychology student. I try to capture the underlying uncertainty of existence in my art. I try to present to people some sort of self-awareness because those were things I thought about a lot growing up in a low-stimulus environment. It’s a thank-you to those who came before me and I’m passing it on.

350 words

What You Need to Know for the APUSH Exam [A Short Note]

Everything’s fairly straightforward. You have 55 multiple choice questions that are stimulus-based. With general knowledge about the trends and important events in US History, you should do just fine. I didn’t study at all for the exam and I was totally fine. That said, that doesn’t mean you shouldn’t study but the exam is going to be easier than the classwork so it’s really not that hard. After the multiple choice, there are four SAQ (short answer questions) where they ask you general questions about time periods. For 2017, one of the SAQs was about the Reconstruction and post-Reconstruction era. If you want an example, APCollege has released the 2017 AP Exam essay questions here. Then, you have a DBQ and an LEQ. Once again, you can see the 2017 essay prompts at the link above. The essays should be the same as the ones you write in class but it’s easier because it’s usually more general and don’t require the amount of specificity that in-class essays usually need.

In conclusion, you’re going to be fine. If you’ve done decently well in class, then you’re definitely going to get at least a 3. If you have trouble, go to your teacher or organise an APUSH AP Exam study group. If anything, you can ask me if you have any questions. Good luck!

Maze Runner: I Made a Game!

TL;DR — Basically, I made a game on code.org for my AP Computer Science Principles (APCSP) class and it’s a maze game and I’ve included the code for it, which can give you a couple of ideas for your own creations and serve as an example, and I’ve also included a video that shows you how it plays and a link for you to play it. There is one big problem with the game code but I’m too lazy to figure it out.

Besides the unoriginal title of this game, I think this little creation of mine is quite decent. There is a problem with the timer compounding every time the game is restarted without the countdown first reaching zero (and a problem which I was too lazy to fix) but otherwise, this simple program is able to execute a game with four screens and multiple game mechanics.

I used AppLab (which uses JavaScript) to make this game and I’ll show you the code used to make it in a second. This was part of the AP project we have to do for AP Computer Science Principles (APCSP) and although I think it’s simplistic considering what you can do on Game Creator, I think it turned out pretty well. In any case, here’s the program code (if you want to use any part of this code in your game, make sure to add a reference to here because otherwise, it is considered plagiarism and the CollegeBoard people doesn’t like plagiarists):

codept1
As always, programming style is important so make it a habit to add comments and indents so that your code is more readable.

As part of the AP project, I also had to make a video with a voiceover since just looking at the program code won’t easily tell you what the game actually does. I also had to make the video under a minute in length so there were some serious limitations on what I could do and I ended up editing the video in a lot of places. Here it is:

 

That’s it. If you want to play the game for yourself, click below and from there, you can see also check out the program code. If any of you guys find a way to fix the timer problem, please let me know.

I’ll talk to you later.

Play the game here!

P.S. I know that the Political section on this blog has been rather empty lately, but I promise we’ll have a new article out in the next couple of weeks. Look forward to it!