New year

Just wishing all a happy new year and add the typical pledge to update the blog more. I just got myself a Nexus 7 to help with this and to be active in the online philosophy community in general.

Will also post some thoughts on using the tablet for academia. So far the biggest lesson is that you need a stylus to annotate PDFs and other texts. Really makes the experience much more enjoyable.

But I will post shortly about the philosophical adventures I've been up to in the last few months.

Not radical enough

Here is a review/critique I wrote on Daniel Hutto and Erik Myin's new book Radicalizing Enactivism.  Like most anti-representation or anti-content theories, they are against content only up to a point.  What I fail to understand, is when such theories introducing representations or content, what happens to all the arguments they just mounted against them?  Why do those arguments only apply to a subset of representations and not when these authors want to use them? For more see below:

Title: A reply to Radicalizing Enactivism: how scaffolding fails to solve the hard problem of content

Extended Abstract: In their new book, Radicalizing Enactivism (RE), Hutto and Myin present compelling arguments for why basic minds do not have content.  In particular, they introduce the Hard Problem of Content (HPC) i.e.: that “informational content is incompatible with explanatory naturalism” (RE, p. xv).  By reviewing a range of theories, the pair demonstrate the futility of recent attempts to distinguish content from covariance.  Content is information within a system, whereas a covariant system can be explained purely by way of causal interactions.  However, in spite of these arguments, Hutto and Myin do not present a theory of mind that is content free.  Instead, they label a total rejection of content as really radical enactivism (RRE), which they claim is a bridge too far.  They conclude that while basic minds do not have content, highly intelligent minds, such as human minds, do trade in content and that this content is made possible due to our linguistic abilities.  In this paper I will challenge Hutto and Myin's conclusion and show how their appeal to language is at odds with the arguments employed by RE to deny contentful cognition.  Hutto and Myin themselves fail to make the choice in the dilemma that they claim HPC presents: either solve how content arises from covariance (even though it does not seem possible to rectify with naturalism) or give up content altogether.

Firstly, I will explicate why an appeal to content is seen as necessary to account for intelligent minds.  For this we must go back to Hutto’s earlier work in which he argues that linguistic minds necessarily have content.[1]  Hutto relies on the Davidsonian model of language, which involves propositional attitudes, such as believing that it is raining outside, or desiring a piece of cake.  These attitudes are contentful because they are about things.  We can know this aboutness because propositions have a truth value.  For example, the statement “it is raining outside right now,” can be true or false depending on the state of the world.  This idea is then used by Hutto[2] to conclude that any language user must have propositional attitudes and therefore content.  In other words, as language is necessarily propositional, any language user must have propositional states in order to be able perform linguistic tasks.

Where Hutto and Myin diverge from a traditional propositional approach[3] is that they insist that the linguistic cognition is only possible through social practices.  This is predicated on the idea of scaffolding, i.e. that the ability to perform certain actions is necessarily linked to an agents' interaction with, and development alongside, elaborate tools and structures in the agent's environment.[4]  This idea of scaffolding appears to be an extremely useful one in terms of understanding the development of human capabilities.

However, what it does not do is solve the HPC, in particular, how the learning of linguistic social practices leads to the creation of content.  This could be contrasted to non-linguistic practices that, supposedly, would not lead to content; for example, learning to use utensils to eat with (forks in some cultures and chopsticks in others).  While there is no doubt that the idea of scaffolding does seem to fit within a naturalistic framework, it does not actually answer the question of how content arises within cognitive systems.  The idea that language is propositional is not validated by the claim that language arises due to social practices.  All it tells us is something about how we learn language, not the constitution of language itself (contentful or not).  To solve the HPC in favor of content, a plausible naturalistic account for the emergence of content is requisite.  At best, Hutto and Myin’s appeal to languages as being contentful could be seen as falling back on the Default Linguistic Mind (as opposed to the Default Internal Mind, see RE, p.137).  This position assumes content is involved in linguistic cognitive practices without reflecting on how it arises within a causal, naturalistic world.

With this assumption exposed, we can see how it works against Hutto and Myin’s particular approach.  For the majority of the book, the authors employ a bottom-up strategy that is at the heart of enactivism: the development of a theory of cognition based on the low-level mechanics of cognition systems.  This is precisely what leads them to conclude that content is often mistaken for covariance and it is covariance that explains cognition at this level.  However, while Hutto and Myin use a bottom-up approach to dismiss other contentful theories of cognition, their statement that language must involve content is driven purely from the top-down received view that language is necessarily contentful.  The bringing together of these two approaches does not make the neat fit that the authors seem to believe it does.

While a bottom-up/top-down switch is common to other enactivist theories, [5] Hutto and Myin’s position relies so heavily on showing that content does not easily emerge from within a naturalistic framework (if at all), the fact that they posit content without specifically tackling the HPC is puzzling.  Moreover, it undermines their critique of other appeals to content: if radical enactivism can simply ignore the HPC when it is convenient, then surely any other theory can too.  The insistence that language must involve content is no different from the insistence that basic minds must involve content if neither side is backed by an account of how content arises.

In light of these concerns, I will conclude that RE presents a strong case for why cognition should be understood as contentless.  The only mistake the authors make is that they view their own work as pushing back the location of content (outside of basic minds), when they should instead view their work as ultimately rejecting content altogether.  Their appeal to language in order to introduce content seems more like a traditional philosophy of language reflex than a thoroughly thought out position.  Only by removing this reflex does the book become a radical work at all.

---

[1] Hutto, D. D. 2008. Folk Psychological Narratives: The Socio-Cultural Basis of Understanding Reasons. Cambridge, MA: MIT Press.

[2] Hutto is not alone in making this step , for a very similar combination of language and cognition see Clark, Andy (1997). Being There: Putting Brain, Body, and World Together Again. Cambridge, MA: MIT Press.

[3] cf. Fodor, Jerry A. (1975). The Language of Thought. Cambridge, MA: Harvard University Press.

[4] Sterelny, Kim (2010). Minds: extended or scaffolded? Phenomenology and the Cognitive Sciences 9 (4):465-481.

[5] Varela, F. J., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive science and human experience. Cambridge, MA: MIT Press.

Thoughts on the AAP conference 2012

I just attended the AAP (Australian Association of Philosophy) conference, the largest philosophy get together in Australia.  It was my second time attending this annual conference, this time at the University of Wollongong.  Even with my limited experience of conferences I think it is a fun but a little weird.  Due to attendance and presentations being open to all comers there is a diverse range of talks and a wide range of attendees.  This tends to make the conference more about getting your field of research across to a wider audience then really tackling a problem in depth.  That said, I have seen very technical talks, and I think that is where I experience the weirdness.  The conference is many different things to different people, I find this makes it hard to connect to people the way you can at smaller and more focused conferences.  At the AAP everyone seems to take pride in butting heads together.

There is also a lot of talk around the big names, seeing what the high profile figures are up to.  This is both in terms of the keynote speakers (which were of a very high quality this year) but also those well-known in the Australian philosophy scene.  For a quick review of those of interest to me: I still don’t understand David Chalmers’ metaphysical project.  The detailed work he does, and encourages his students to do, is a level of technicality I don’t see as being beneficial.  The main concerns around his work don’t seem to got away anytime soon through careful metaphysical considerations.  That said, you cannot fault the quality of the work.

I have a similar concern with Peter Godfrey-Smith’s work.  In his talk he discussed the topic of information transfer between animals.  It must be said, it was a fascinating topic and the insight provided into a bias in Dawkins work was informative.  As was Godfrey-Smith’s appeal to David Lewis, which will no doubt offer something useful to the topic.  Yet the overall treatment was very black-box in nature as “signaler” and “receiver” were simple components in a very simple model.  This is at odds with the modern embodied-embedded/ecological approach.  If there really is a revolution towards more holistic approaches it is a very slow going one, having to reach each topic in philosophy individually.

While I was not able to attend I heard that Kim Sterelny’s work is still impeccable, probably the best at tackling philosophy of mind from an evolutionary/biology perspective.  Also the talks by the La Trobe contingency went well, though there seemed to be a strong accented voice of disagreement in a couple.

My own talk was on at a bad time, the first talk of the morning the day after the conference dinner.  So not many people had yet recovered.  I was told it was good by those that did attend, basically because everyone agreed.  This type of feedback is not the best but it is nice change to the normally combative comments I get.  The talk was essentially the same challenge to affordances that I outlined in my previous blog post.  From the mammoth text below I changed it into a more presentable form and tighten the argument.  The conclusion is still the same, that affordances can explain innate perception to action behaviours but not more complicated cases.  For that, something more is needed.  Whether it is representations or something else I left it as an open question.  That is the work of the next chapter in my thesis and will be revealed soon.

As a last comment both on my talk and the conference: it is only when I go to such conferences that I realise how “heideggerian” I am in contrast to many of those working in philosophy of mind.  This both excites and scares me.  For on one hand I have a good amount of knowledge that many do no posses, while on the other, the work I do seems outside the radar of most.  However, I think it is too late for me to change now.  I both enjoy this type of philosophy too much and also think it is right way for phil. of mind to be heading.  All that said, the conclusion at the end of my talk was that Heideggerian AI was Heideggerian enough.  I regret not using that as the title for the talk itself.  Maybe next time.

How much do affordances explain?

The concept of affordance has become an important component in explaining the nature of perception.  The idea of affordance was introduced and developed by J.J Gibson as a radical reinterpretation of perception, one that champions direct perception.  Instead of saying that we perceive the features of the world around us, and from this construct a model of what we are perceiving (indirect perception), Gibson suggested that we directly perceive possible actions in the world.  So a flat surface affords the possibility of walking upon it; a small rock affords the possibility of being picked up and so on.

In contrast, the perception of features involves detecting the property of flatness, stableness etc., to then represent these properties and from that make an inference that the surface can be walked upon.  While the inference model is still the most popular theory in terms of explaining cognition (the representational theory of mind), there is a renewed appeal for direct theories of not just perception but cognition in general.  This appeal is driven by the simplest of ideas: in order to perform day-to-day tasks we don’t interact with objects in our head but with the world directly.  There are also several arguments against the the inference model (especially mental representations) that also give direct cognition more credence.  However, like many, my main reason for being attracted to the theory is that there does seem to be something direct about a lot of our cognition.

Even with that said, I am not a full blown supporter.  In short I think the conclusions that are derived from this claim are often too strong; taking the idea too far without extra steps of justification.  This doubt comes through in the use and explanation of affordances.  Specifically when we ask: what types of information do affordances give us access to? In the most sloppiest of uses, the concept of affordance becomes synonymous with direct perception and in this way fails to offer an explanation of what affordances are.

Before discussing the criticisms, I want to point out the cases where affordances are a constructive and useful idea.  There is a good reason that the walkability (affordances are normally talked about the abilities they present) of a flat surface is the de rigeur example introduced by Gibson.  If we consider such things as new born foals being able to walk soon after birth this seems to say something about their ability to use their perception to know where they can and can’t move about.  Contrast a foals movement with a newly hatch spider and its freedom to climb vertical surfaces.  Or consider how new born turtles use gradient and light to guide them towards the sea.  To say that the turtles use the light and gradient to deduce the position of the sea seems to create an overly complex model, one in which the turtle needs time to pause to think.  The idea of affordance helps remove this extra layer of complexity.  For the baby turtle to see the light is to see not the world but the instruction: move in this direction (or perhaps the more catchy: walk this way).

This does not mean that the turtle has some semantic content about movement, rather it is the idea that the light is a cause of its movement not a step in a long process.  The ambient energy perceived by the turtle is meant to activated the actual perception of possible movement, not some secondary semantic content.  This also gives insight into the nature of direct cognition.

This idea of directly perceiving actions seems a plausible explanation for such cases, namely cases of innate or instinctual behaviour.  While representationalists can still argue that there is the possibility that inferences are occurring even in these instinctual cases they have other concerns they must overcome, the most talked about currently being the frame problem.  My concern, however, is though we can accept affordances working at this innate level, does the concept extend further?  What about behaviours that are learnt, ones in which the association between perception and action is not always so straightforward?

Here I am thinking of cultural practices, consider the use of chopsticks or a fork to eat with.  Is there anything about a fork or pair of chopsticks that goes from them being perceived to being used to put food into a person’s mouth?  Both might be perceivable as being pick-up-able, but does this then lead to put-food-in-mouthable?  If affordances are directly perceived it cannot be claimed that that to see a fork is to make an association that this is a tool to eat with, rather, to see the fork should mean an actual perception of using the fork to eat with.  However, for those unfamiliar with forks, or with chopsticks, I think there is no such immediate perceptual act going on.  There instead seems to be an association between what is perceived and how it can be used.  With theories of direct perception there seems to be no room for a mental thing that makes such association possible.

To step back a little, the idea of affordance claims that when an organism perceives a particular feature of the world, that very act of perception reveals the affordance.  This idea was challenged by Fodor and Pylyshyn (1981), arguing that the ambient energy that we interact with (ie the stuff that engages our senses, like light rays or sound waves) must only carry information relevant to its medium.  Light rays carry vision specific information, sound waves, hearing specific etc.  This was shown to be a faulty argument by Turvey et. al. (1981) in that the Fodor and Pylyshyn did not give a great enough role to the act that perception plays in an affordance.  Affordances are a union between an organism (organic or not) and the world.  That organism does not just perceive like a video camera but has needs and ways to move about to get those needs.  In his description of an affordance Gibson stated an affordance must be understood as the coming together of the needs and ways to move of an organism with ambient energy from that organism’s surroundings (Gibson 1979, p. 129).  Affordances need this union between world and organism in order for them to avoid Fodor’s criticism.  For this reason I think those that claim that affordances are ontological independent of this union are mistaken.

However, even with this clarification, we can see why learnt behaviour is such a problem.  If affordances are the relation between the ambient energy and the organism, then what changes when a behaviour is learnt is only the organism.  The ambient energy remains the same.  Nor, can it be said that the perceptual capability of the organism has changed (there are a handful of exceptions to this, namely being able to actually disambiguate ambient energy, as when seeing through a microscope).  When I personally learnt to use chopsticks it was not due to some perceptual awakening, they did not glow or present themselves in a new way.  Instead, I can now associate the shape with the act of picking up food and eating even though the shape of the object was one I was able to perceive before I knew they could be used to eat with.  In this way learnt behaviour is about the change that the cognitive agent has gone through.  And these changes do no involve the a change in the nature of perception but a response to a perception.

At this point I suspect those wanting to extend affordances to learnt behaviour will appeal to some type of dispositional based behaviour.  That by developing different types of responses to certain environmental conditions, learnt behaviour can still be explained by affordances.  Here I agree that the ability to deploy certain behaviours in particular circumstances does offer some way forward in solving this problem.  However, as I will argue another time, it isn’t enough.  Just like the idea of affordance, dispositions emphasis the link between the state of the world and possible actions that can be taken.  And just like with affordances there is still no additional cognitive mechanism to account for association.  Thereby falling to the same criticism presented here.

This problem I find especially interesting if one rejects the temptation to fall back on representations.  How is it that perceiving something in the world allows for not just one act of learnt behaviour to be considered, but also allows for a raft of possibilities, ones that we may never have even encountered before (for example, a fork can, in hopefully rare circumstances, quickly turn from a tool to eat with to a tool to injure with).  As it stands I think affordances tie the relationship between world and organism too closely together.  There needs to be more space for cognitive agents to develop and influence their own behaviour.

Not really that related but this comic:

http://www.smbc-comics.com/index.php?db=comics&id=2543

Got me thinking how the act of perceiving a blue colour wave length as red or green because it is an apple is a common way to show that our ability to perceive is an action.

The act of intentionality in this case is the drive to see things as similar, our perceptual system wants to keep the perception of apples consistent and to do so would require the applying of some act, some instruction or rule in order to do so.  From this we can state that to perceive an apple, in the very least, requires some cognitive act in order for it to have this consistency.  From this most would want to go from distinguishing mind driven content to world driven.  But instead, I want to think in terms of different acts that we perform, some more consistent with the world around us than others.

Anyway, it seems the same problem is still floating around in my head.

2 types of intentionality

To start off this blog I want to explore an important distinction that I seem unable to get away from in my own work on mental representations.  While it is often presented in its Heideggerian form, that of present-to-hand versus ready-to-hand or his ontic/ontological distinction, I want to approach this distinction from a different angle and talk instead about two types of intentionality.  The first type of intentionality is the more traditional way of interpreting the work of Brentano, that our thoughts are directed towards something, an object.  This is often equated with the idea of our thoughts being about something, so I am thinking about the current Australian cricket captain and his constant back injuries.  There is a question of whether aboutness is really the same thing as being directed towards something but in both cases the idea is that our thoughts are about objects, abstract or real (for example we can think about the whole Australian cricket team as one thing, yet it clearly is not a singular object).  This type of intentionality purposefully relates our thoughts to the attributes or properties of the objects in the world around us, suggesting that our thoughts are about always about objects and their properties.  For this reason I will call this the intentionality of objects.  This type of intentionality is generally associated with the computational theory of mind.

This is in contrast to the intentionality of actions, in this case the focus is not on the objects in the world but rather the task and actions we carry out.  Here the suggestion is that we do not fill our minds with objects and their properties, rather, the content is about what we want to get done.  This is often related to everyday tasks, for example, when I go to the supermarket and select the food I want to buy, I do not select broccoli due to its properties of greenness or high iron, rather I select it as it goes into the stir fry dishes I like to make.  My thoughts are always on the role the item would play in actions I want to perform.  This is true even if I consider what broccoli is made of, for example, how it is made of healthy biological substances such as iron.  In this case the intentionality of action is to lose weight or to be fit, rather than the pure assessment of properties of objects.  This action based approach is associated with the current anti-representationalist camp; those that deny that representations play a part in our mental or cognitive acts.

This discussion is one that I am heavily involved in with my PhD thesis but instead of focussing on the computational versus the task based approaches (which is the common approach in cognitive science studies), I want to examine the topic from within a philosophy of mind framework.  Specifically through the the question: can we say that our thoughts are about objects or actions?  There are three obvious responses we can have to this question:

1) Thoughts are about objects not actions
2) Thoughts are about actions not objects
3) Thoughts are about both actions and objects

Heidegger is a supporter of 3), he states quite clearly that he thinks that we can switch modes of being, each mode allowing us to think about objects or actions but never both at the same time.  Computationalists would disagree, stating that our thoughts always have some object filling them.  For example, when I think about myself running I do not think of some pure running-ness, rather I direct my attention to particular objects of thought, such as the footpath I run on or the dog I encountered along the way.  This makes them very much in agreement with 1).

What I find interesting is the question of whether there is a case for supporting 2); if you can argue that our thoughts are always about actions.  This is against the much more ingrained (in analytic circles at least) approach of 1) or the broader approach of 3).  Where I see the potential for an argument for 2) is in our changing attitudes towards perception.  The idea that we do not sit back and simply take in the world through our senses is something that is well discussed and generally accepted to be true.  Alva Noë, for one, attempts to link this understanding of perception as an action to argue against traditional object based theories of mind (see his book Action in Perception 2004).  When you claim that our acts of perception are not this passive way of revealing the world around us but are instead active, engaged tasks that people carry out it seems that the work for supporting 2) is done.  We just need to understand that the objects we see in front of us are not some reproductions of reality but rather constructs we have created in conjunction with external stimulus to make a perceptual scene.  However, even though this type of argument is not new and is well known, this idea seems a conceptually difficult one to respond to; we seem reluctant to change our deep held view of us as neutral perceivers of the world.

This leads to the puzzle that I have been trying to solve once I became aware of this approach.  Essentially it is trying to come up with a way to understand perception as being an action.  The problem here is how do we understand our perception of the world to be an action (or set of actions)?  We so easily see our bodily movements as actions due to their observable status yet this is not the case for anything happening in our brains.  Also, our brains don’t make these clear precise actions that we associate with bodily movement (a kick, a punch).  The act of perceiving appears to be temporally stretched.  There is no clear delimitation between making one perception and another.  So how is it that we can understand something like perception as being an action?

Oddly enough, the slight insight I have into this problem comes from theories of computation, specifically how computers work.  When any a piece of data is brought to the CPU of a computer, there is always some task that is carried out upon it.  A binary value is added or compared to another, the data may simply be moved and so on.  The reason for bringing a piece of data to the CPU is so some action to be carried out using it. So how could the act of perception be like this type of computation?  The example that I keep returning to is understanding the perception of colour as being founded on the act of comparison.  Seeing something as red as opposed to blue helps us mark out two differing things in the world.  Two things we may want to engage with in different ways.

This idea that colour perception is an act of comparison is fuelled by the way we see consistency in colour when, technically, there is none.  We see a whole wall as being white in colour not because it gives off the same lightwaves at all points, rather, it seems we see it as being one colour as it is we do not want to distinguish it into different parts.  There is a holding back of comparison in an effort make some things the same while others different.  What makes for an interesting question is that are we as perceivers experiencing colour or experiencing comparison?

This, of course, is not a complete answer for seeing perception as action.  However, I think it is a start.  As I mentioned this is something that is not particularly intuitive so I am content with small ideas such as these.  Hopefully, as a side topic from my main project it is one that I can return to at a later date or hear more about from others.