18 Temmuz 2015 Cumartesi

Being in the world

He threw the immaterial Cartesian soul into the world and defined him mostly and profoundly as a being in the world (Dasein). He broke the epistemological bridge between the subject and the object, and literally ended the Western metaphysics as we know it. Nothing would remain the same after him. I am talking about Martin Heidegger, who was arguably the most influential philosopher of the 20th century. This blogpost has too little space and I am too mortal and naive to go too deep into Heidegger’s philosophy, but I will survey the basic insights of his work that are still resonating with us today (Thanks to my good friend Larry for recently pointing out Heidegger’s significance for metaphysics that gave me an inspiration for this post).

I sometimes wonder how philosophy would develop if Descartes never existed (no pun intended). Much of the philosophy came after him was shaped by his ideas and many legends were born through their rebellion to Descartes. Ever since Descartes philosophers were stuck in this question of knowing about the existence of an external world. Following the old tradition of subject/object ontology, one of the number of bad legacies Descartes left was defining the subject as an immaterial soul (res cogitans) that is separated from the world and the body (res extensia). Is there a world out there that exist independent of us, or all we can access is our representations of a world? Immanuel Kant called it a scandal that this question had not been answered before. But Heidegger said the real scandal was the very attempt to pose this question over and over again. This was the fundamental ontological break of Heidegger from what preceded him. Heidegger said posing of this question left the very investigation the word “to exist” untouched. He pointed a very simple question that surprisingly went unasked throughout history: What is this being we are talking about, e.g. what is this thing called sum in ‘cogito ergo sum’? Heidegger took up the challenge to answer this question, and the result was a formulation of a whole new vocabulary that culminated in a phenomenal and yet quite difficult book called “Being and Time” (1927).

Heidegger said we were essentially coping with the world rather than setting back and thinking about it. He gave his famous example of hammering.  In the act of masterful hammering there was no Cartesian split between the subject and the object. There was just hammering without reflection. Subject melted in the activity, the division between a subject and object disappeared (Merleu Ponty would later call this “skillfull coping” with the world). Now I am not a masterful carpenter so perhaps I would be too Cartesian in the act of hammering but you can think of other examples which involve such ‘procedural acts’ such as driving, opening the doors, playing guitar etc. These Zen-like smooth coping with the world were our fundamental mode of activity (Given this priority of the ‘bodily’ actions, it is so strange that Heidegger did not elaborate on the body in Being and Time- a project left to Maurice Merleu Ponty).

 First and foremost, being (Dasein) was in the world and always grounded in the world. Being was fundamentally temporal: it was thrown into the world that he did not choose (past), always participating in the world as a skillful coping (present) and fundamentally open to the possibilities (future). And being always had an attunement (mood) in everyday activity. His mode of being always resonated in a certain emotional frequency, constantly shifting and drifting through time. This projection to the future allowed the being to face the reality of his own death and this knowledge of it’s own finitude was the fundamental source of meaning of it’s existence (a link to existentialism).

 I will here focus on the extent his ideas resonate in the field of cognitive science. Heidegger crashed the subject/object ontology and the Cartesian tradition, but ironically cognitivism had it’s prime years long after Being and Time was published. Cognitivism, using the computer metaphor and the Cartesian subject/object ontology, basically claimed that what we call thinking is a symbol manipulation based on the stored representations about the world. You can see the Cartesian link here: there is an epistemological barrier between the subject (mind) and the world, and the only way to generate cognition is taking input (sensory information) from the world, doing bunch of manipulations to them according to certain algorithms, and then producing the output (behavior). Cognitivists working in the field of artificial intelligence were really hopeful that they would produce human-like robots within few years. Now in retrospection it would be fair to say that this ideal failed completely. It produced some success in purely abstract domains like chess but the robots were and still are far from skillfully hammering nails or navigate their way smoothly in the world. Why? It was not because the machines processing power were not good enough. It was because of the following foundational mistake: abstract, ungrounded and detached model of cognition simply did not come close to how human cognition actually operated in the world. The body was neglected.

It was another influential philosopher, Maurice Merleu Ponty who carried Heidegger’s ideas to the next level and brought the body into the picture. In his monumental book “The Phenomenology of Perception” (1945), he described how body and perception allows a skillful coping with the world and this process was the foundation of all cognition. Similar to Heidegger’s example about hammering, Ponty pointed to the fact that our bodies allow us to increase our grip to the world in the act of perception: when we are looking at a picture in an art gallery, we move our bodies to an optimal distance to the picture that allows a meaningful interaction with it. Perception is fundamentally a sensorimotor process and following Heideggerian tradition, there is no intermediary between perception and action. No symbol manipulation, no representation. Sensing and acting are fundamentally coupled to each other and the action flows from perception in a dynamic process involving the body and sensorimotor structures.

Ponty’s ideas did not go unnoticed. Many failures of the cognitivism framework was calling for a paradigm shift. The shift came with the new framework that is called embodied cognition. Especially Gibson’s work on affordances was influential to lead way to this new framework. In a nutshell, instead of talking about symbol manipulatins and representations in the brain, embodiment researchers were articulating the tight coupling between body, environment, history; and cognition was generated from the historical process of organisms bodily interactions with their environment.

A very influential book called “The Embodied Mind”  (1991) pointed a link between Heidegger/Ponty and the embodiment framework. They also articulated how buddhistic teachings were much more compatible with the new framework compared to the western metaphysical tradition. According to those writers a fundamental paradigm shift comes from the realization that there is not a fixed Cartesian ego to be found. The mind is fundamentally without a ground. It’s a dance of interaction between the body and the world that we call cognition. Thoughts and experiences are always fleeting, they are like clouds coming and passing. There is no ‘owner’ of cognition. It’s all a process. Even memories do not stay the same, they are constructed each and every time of recollection, and just as you cannot swim in the same river twice, you cannot have two identical memories at different times.

Why did Heidegger only grant Being (Dasein) to human beings? Why did he exclude animals? Why did he never talk about evolution of Dasein? Knowledge of one’s own death could be one of the reasons. Knowledge of it’s own existence could be another. But we have to answer a simple question: At which point during evolution the being and meaning emerges? Can’t we claim that animals are also sense-making creatures with their own sensorimotor interaction with their environment? Of course they are lacking general knowledge about the world and most of them are not pondering about the meaning of life like we do, but they have bodies, moods, memories and foresight abilities. They are skillfully coping with the world and are much better at that compared to the most intelligent robots today. Today many subfields are dealing with this question of evolutionary meaning generation, such as biosemiotics. They claim that meaning and semiosis (interpretation of signs) constitute intrinsic parts of life processes. Thus we can talk about meaning even in life of bacteria: it can interpret glucose as good and swims towards it while it avoids other chemicals.

Going back to artificial intelligence (AI), the field of robotics at last accepted the failure of cognitivism and they are now using the embodied approach to build robots that learn about the world through their bodily interactions. However, they are not ‘Heideggarian enough’ according to the famous Heidegger scholar Hubert Dreyfus. In his 2007 article he suggests everyone working in this field should study Being and Time but he is still pessimistic when it comes to achieving a human-level intelligence. He claims this would fail due to the problem of significance such that we assign significance to the objects in the world depending on our specific needs, cultural backgrounds and bodily structures. In order to solve this issue, Dreyfus offers the following far-fetched recommendation: “… we would also need- and here’s the rub—a model of our particular way of being embedded and embodied such that what we experience is significant for us in the particular way that it is. That is, we would have to include in our program a model of a body very much like ours with our needs, desires, pleasures, pains, ways of moving, cultural background, etc.” Does it sound like a tenable implementation? It does not seem so.
Of  course many AI researchers do not give up so easily. They claim that machines are already more intelligent than humans in many domains and they could be excused for not being able to boil a cup of tea. In this video one of the AI researchers mentions a scary scenerio (around 05:30 min.) where the military missile system guided by robots is attacking us and we are making fun of them because they can’t make a cup of tea. This is precisely a scary scenerio not because machines can’t make a cup of tea but because the machines lack of normativity could be used badly in evil hands. In Heideggerian terms those machines are not Dasein, they are just zombies.
P.S. I want to recommend a great radio show called ”Entitled Opinions” hosted by Robert Harrison. He is a big Heidegger fan and the intellectual quality of the show is enormous. Secondly, you can watch Hubert Dreyfus’ full 28 lectures on Being and Time here.
Can

References

Martin Heidegger, Being and Time, (1927). Oxford: Basil Blackwell (1962).

Maurice Merleu Ponty, Phenomenology of Perception (1945). Colin Smith New York: Humanities Press (1962).

Eleanor Rosch, Evan Thompson, Francisco J. Varela (1991). The embodied mind: Cognitive Science and Human Experience. MIT Press.

Hubert Dreyfus, (2007). Why Heideggerian AI Failed and how fixing it would require making it more Heideggerian. Philosophical Psychology, 20 (2) (2007), pp. 247–268.

Gibson J, (1979). The Ecological Approach to Visual Perception. Houghton Mifflin, New York.





13 Temmuz 2015 Pazartesi

Escaping from future whippings and electric shocks: Prospection unites us all

Here is a true story: Three little cousins are playing games. In a sudden burst of aggression, the youngest one throws a slipper to the eldest one’s face and slaps him. Eldest one starts crying. All of a sudden the middle one starts crying too. Family members are surprised and they ask him why he is crying. He replies: “What if he beats me too?” What a remarkable skill of prospection! This post will be about this notable feat of anticipation.

We all know that future doesn’t exist, right? How can a nonexistent thing can be a cause of anything then? This seems to go against the rules of physics. In fact, the very involvement of the concept of ‘future’ in a scientific explanation seems to resurrect the long abolished concepts of teleology and Aristotelian final cause (that for the sake of something happens, the purpose of an action). Indeed, as Seligman put it elegantly in his recent and influential paper, 20th century science was purely mechanistic: it asserted that the behavior is completely governed by past causes (Freud) and animals do not have expectations about the future but they behave, or rather react, only according to reward-punishment contingencies they experienced in the past (Skinner- behaviourism).

According to behaviourism, animals were not capable of expecting anything about the future as they were mere machines following a very simple rule: repeat those behaviours that rewarded you in the past and banish those behaviours that punished you in the past. However this picture has gradually changed thanks to the very own results of the behaviourism practice. Rats were using novel shortcuts to reach the rewarded sites as if they had a cognitive map in their brain. It was as if they were calculating alternative roots before ever experiencing them. But here was the challenge: what sort of tests could distinguish the animals acting according to future expectations from those merely reacting according to the history of reward and punishment? One such evidence came from an avoidance learning experiment that is cited in Seligman et al. (2013). Now avoidance or escape behavior looks like an ultimate future-oriented behavior, isn’t it? “No!” said the behaviourists! They claimed that expectations were not playing any role in the causal chain. And they thought their explanations reduced such seemingly future-oriented behavior to mere reward/punishment contingencies governed by the past. Were they right?

The experiment involved many rats, a tone, an electric shock, and a barrier that rats could jump over to the safe side. Rats were first given experience where the tone was followed by an electric shock 5 seconds later. They could escape the shock by jumping over the barrier within those 5 seconds. Their jumping also stopped the sound of the tone. After some time all the rats became steady jumpers. The crucial question was this: Did rats jump to prevent a future electric shock or did they jump in order to stop the sound of a tone that was associated with electric shock during training? Cognitivists chose the first option while behaviourists picked the second one. Behaviourists claimed that during training, the tone became an aversive stimulus through Pavlovian conditioning processes as it was followed by an electric shock. Hence, rats were escaping because this behavior was rewarded as it caused the tone (aversive stimulus) to stop. Notice that there is no mentioning of any notion of expectation in this scenario. Conversely, cognitivists claimed that rats were escaping to prevent a future electric shock.

The critical test came in the later phase of this experiment. If behaviourists were right and rats were jumping to stop the tone that led to punishment during the training trials, then you would expect the rats to stop jumping after hundreds of trials because it would lead to extinction: the tone would lose its aversive nature as the it was never followed by shock throughout those hundreds of trials. However if cognitivists were right and the rats were jumping in order to escape a future shock, then you would expect them to keep on jumping after hundreds of trials as this “if-then” conditional would not be broken by extinction processes. (If I jump, I will escape a future shock. If I do not jump, I will get a future shock). Result? After hundreds of trials, rats were still jumping! This was a victory for the cognitivists and a defeat for behaviourists, who later modified their theories to give expectations an integral part to explain the learning phenomenon. Nowadays neuroscience is telling us that rats are engaging in mental time-travel as their brain regions representing a room get activated at the choice point before they enter into that room. (Johnson & Redish, 2007).

Today it seems like we are leaving behind the old notions of behavior is governed and completely determined by the past, and that we are imprisoners of those past events that we have no control of changing. Today we are more future-oriented: the very roots of action, motivation and meaning are densely intertwined with our representation of the future. This reminds me of the Heideggerian notion of “throwness” as an integral part of Dasein- future projection is an inherent characteristic of a being in the world.

In fact, notions such as expectation, prediction, anticipation and prospection become very fashionable in cognitive science. Now we all know that the value of predictions in everyday life. For example we cannot tickle ourselves and the crux of a joke is its surprise value, i.e. how much it deviates from our expectations. One might say driving is all about anticipation: you have to constantly predict the behavior of the drivers and take action accordingly. Similarly goalkeepers in football have to predict ball’s trajectory to make a save. But one recent theory makes a more radical claim: it’s all about predictions from neurons to behavior!

Here is the premise of this theory: Brain is not a passive machine that is driven by outside stimuli; on the contrary, brain is very pro-active: it constantly throws predictions on its perceptions, it looks at the world with expectations and constantly chisels the sense data to fit into its expectations. In fact, brain does not process ‘already predicted and expected’ perceptions. What is worth processing are so called ‘news’, those information that are surprising for the brain, namely those sensations that do not fit into prior expectations (Clark 2013). This surprise element constitutes the core principle of this theory and has ramifications in similar theories like the Bayesian brain hypothesis and free-energy principle that I don’t have space to delve into (Friston, Stephan, 2007). But the basic idea is that the brain needs to eliminate the surprise either by changing the prior expectations or changing the very perception to fit into those expectations (hollow-face illusion). This theory becomes even more radical when it claims exactly the same processes are in play when it comes to action: When we act, we predict the future state of the world and then act to “fit” the world to our expectations. This theory hence ambitiously offers a continuity between perception, action, imagination and planning.

I find this aspect of the theory really interesting as it points to the goal-directedness of the actions. Since we cannot change the past, all the possibility of change resides in the future. Action is taken in the present moment in order to reach that desired future state.  These ideas were in fact articulated in the beginning of the 20th century. William James said that all thinking is for doing, and since doing can only change the future and never the past, cognition can be said to be ultimately prospective. Similarly when Henri Bergson said “the past is powerless whereas the present is sensori-motor, and therefore active”, he was depicting a similar prospective picture for the mind.

We will see how further developments will change the picture. One important thing is that even if cognition is fundamentally future-oriented, this does not mean that we should expect long-term planning in other animals that involves a detachment from the needs and motivations of the present moment. After all, loosing oneself in the imaginary future scenarios can be a costly activity, both in terms of cognitive resources as well as the vulnerability to the possible dangers in the current situation. After all, world is a dynamic and ever-changing place and daydreaming might have substantial evolutionary costs for many species.

Going back to those cousins again, the middle cousin that cried in anticipation of a future slap was yours truly, who was arguably a very cautious child with remarkable avoidance learning skills. Perhaps that’s why he is doing a PhD. on prospective skills of animals J

Can


References

Seligman, Railton, Baumeister, Sripada 2013. Navigating into the future or driven by the past. Perspectives on Psychological Science 8: 119-141.

Clark, A 2013. Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioural and Brain Sciences 1-73.

Johnson, A., & Redish, A. 2007. Neural ensembles at CA3 tran- siently encode paths forward of the animal at a decision point. Journal of Neuroscience, 27, 12176–12189.
James, W. (1890). The principles of psychology (2 vols.). Cambridge, MA: Harvard University Press.
Friston, K. & Stephan, K. (2007) Free energy and the brain. Synthese 159(3): 417–458.
Bergson, H. Matter and Memory, tr., N.M. Paul and W.S. Palmer, New York: Zone Books, 1994.



10 Temmuz 2015 Cuma

Processing, please wait...

There was so much wisdom in pre-Socratic philosophers. My favourite one is Heraclitus from Ephesus, who lived in modern-day Turkey and made into history with the following two-word claim: “Panta rhei” (Everything flows). This was a foundational ontological claim about the nature of reality as for Heraclitus there was no immutable substance at the core of reality, but there was constant transformation, flux and change. This idea was in a stark contrast to the ones proposed Parmenides and Democritus. According to Parmenides the change was impossible and existence had a timeless nature and for Democritus the everything was made up of indivisible particles (atoms). Hence, for Heraclitus it was all about process and becoming whilst for Parmenides and Democritus it was substance and being that constituted ontological priority.

Plato and Aristotle liked the ideas of Parmenides and Democritus more than that of Heraclitus and the western thought was found on substance ontology and atomism. This essentially means that at the core of things world is made up of stuff, substance, indivisible particles, and the transformations and changes these particles undergo have only a secondary nature. As Nicholas Rescher puts it, substance ontology is “subject/object” oriented where the agency is attributed to the subjects/objects in the world, doing things. This stems from Aristotelian logic and culminates in Descartian res cogitans and eventually in modern western science. However, process ontology claims “becoming” is no less fundamental and more pervasive than “being” and it does not necessarily entail an active agency from a subject. As in the case of water evaporating, we do not attribute agency to water molecules, such that they “do” the evaporation. And yet the world is full of such processes happening all the time. But our minds are biased to attribute agency to all sorts of actions. It’s highly bound to our causal reasoning that leads us from effect to causes. We are hardwired to look for causes in the world, and most often subjects are responsible for those causes; e.g. a lion leaving tracks or even a divine intervention into the world. Our minds have to abstract away and fixate certain regularities in the world, even if those seeming regularities undergo a constant change. That’s why it is intuitively so hard for us to accept the process ontology, as we feel like the very ground we stand on is taking away from us, we find no place that we can cling onto, everything is a one constant flux.

Here I will briefly mention the link between Buddhism, self (ego) and process philosophy. Buddhism doctrine embraces the idea of constant transformation in the universe, and claims that the root of the psychological suffering is the strife to desperately cling on to a fixed entity like self or ego when there is none. While in the west we tend to see the self as a firm locus directing our actions and giving us a sense of agency, eastern theology sees this self as an ever-changing entity, constantly in a process of becoming and self- transforming. Even David Hume, a prominent western philosopher, articulated this labile nature of the self: “For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception”.

For Henri Bergson, language was a burden to capture the flux of reality. The very nature of the language transforms the ever changing nature of reality into dead words and concepts. Then we confuse reality for the words. The very tool of communication we invented is a burden in understainding the true nature of reality, which is in a constant process of change. He put it beautifully here: ”We confuse the feeling itself, which is in a perpetual state of becoming, with it’s permanent external object, and especially the word which expresses this object… not only does language makes us believe in the unchangeableness of our sensations, but it will sometimes decieve us as to the nature of the sensation felt.” For Bergson, there is no possibility of two conscious states of mind that are identical to each other for the latter already involves the memory of the former- this induces the fundamental nature of temporality into the equation- as opposed to physics, irreversability of time is a fundamental aspect of life and mind. The time-bound developmental trajectory can create novelty that is not predicted by physical laws which does not differentiate between the past and the future.

 In fact, Bergson was a hard-nosed opponent of Einstein and he criticized Einstein’s treatment of time in his general theory of relativity as an observer-independent, objective entity that could be measured by clocks. For Bergson, this treatment of time took away the meaning of lived time we all intutitively experienced in our lives. Bergson conceptualized this temporal aspect of lived experience as ”duration”, and he claimed that no image can replace the intuition of duration. Obviously, Bergson’s fuzzy concepts and definitions did not impress Einstein who had hard times understanding him. In a historical encounter between these two most prominent minds in 1922, Einstein dismissed Bergsonian time in a single sentence: ”Time of the philosophers does not exist.” Later Heidegger would take sides with Bergson in his monumental work ”Being and Time” where he depicted Dasein as a fundamentally temporal subject, and argued that time should be understood as a process (becoming) rather than a divisible thing or an entity, like space.

Processes are ultimately temporal- they unfold over time and display chronological coherence. Thus process ontology might be especially useful when it comes to complex phenomenon like life and mind as they are time-bounded and historical domains and they involve many components and subcomponents interacting with each other. In physics you can obtain universal laws and equations that are applicable to various domains. However when you ascent in scale and reach biology or psychology, general laws do not seem to apply and you have to give up truth for simplification and abstraction. It’s called Bonini’s paradox and goes like this: ”as a model grows more realistic, it also becomes just as difficult to understand as the real-world processes it represents"

Terrence Deacon recently attempted to explain the phenomenon of life and mind by using process ontology and the concept of emergence: according to his account higher level processes can emerge from synergistic interactions from lower level processes. While the stuff those processes operate is constantly changing and replenished, the form stays the same thanks to the organization of those processes- reminding us Heraclitus’ river argument: even if the river form is preserved, the molecules consituting the river is constantly changing and replenished by new molecules. Same analogy can be applied to our bodies- our cells are constantly dying to be replenished by new cells. Similarly, our minds are constantly changing as we accumulate new memories and learn new things. Time also manifests itself in an evolutionary scale: for millions of years in this planet there was no life or mind; but now they are pervasive phenomena in the world. Is there a certainity that the events on earth would unfold just the same way if we replayed the tape of life? Probably not. This stresses the enchanting aspect of evolution and time: it’s ultimate openness to novelty and contingency.

Process philosophy can also be helpful in avoiding homunculus fallacy that often arises in scientific explanation of cognitive phenomena. Homunculus literally means a little man, and homunculus fallacy denotes attributing agency to this little man as an explanation to a phenomenon but this leads to an infinite regress as there must be another little man inside the head of this little man and so on. Typical example of this is in the realm of vision, as a little man in the back of the skull is depicted to ”look” at the images formed on the retina by the light rays as an explanation of how vision works. The fallacy arises as this explanation just begs the very question it claims to address: how does that little man see? Since homuncular explanation is no explanation, we must explain the process of vision without attributing an atomistic subject of agency: hence there is no one seeing, there is only action of seeing; seeing just happens. This leads us to an important point: these dynamic cognitive processes can only be understood in ”action” terms because they are in a constant process of flux, change and interaction. Clinging into the substance ontology and vocabulary might not capture the real world dynamism of these processes as cognition is essentially about the ’transition’ between mental states rather than the mental states per se.

We have a very daunting task ahead of us as cognitive scientists. Cognition represents the penultimate stage of complexity in its ambition of answering the question of how the brain works. What I suggest here is to re-consider our ontological approach to the field and take into consideration it’s process nature. Let’s end with Henri Bergson: ”In the human soul there are only processes.”

Can


References

Nicholas Rescher, 2000. Process Philosophy

David Hume, 1740. A Treatise of Human Nature

Henri Bergson, 1888. Time and Free Will

Martin Heidegger, 1927. Being and Time

Terrence Deacon, 2011. Incomplete Nature: How Mind Emerged from Matter