JellevanDijk.org

Interessant stuk van Dimitri Tokmetzis over een documentaire van Adam Curtis over de relatie tussen mens en technologie (met als uitgangspunt dat de technologie die we maken ‘onze’ natuurlijke wereld is waarin wij leven. Die dus niet natuurlijk is. De discussie over hoe we onszelf zien (als een soort van complexe biologische technologie) deed me denken aan een boek van Douwe Draaisma: de metaforenfabriek. (Al heb ik het idee dat het vroeger anders heette en een heruitgave is, maar zijn andere boekeng aan over de faalbaarheid van het geheugen dus ik vertrouw mezelf voor geenmeter)

Popularity: 10% [?]

O no. They’re doing it again. There are people discussing whether or not the “Google car” is conscious. Apparently nobody stopped them short. It brings back good old memories about good old artificial intelligence, and all the moderner types (connectionism, Alife, behavior-based robotics, etc…).

So I was thinking, we can either answer this question from an ‘engineering’ perspective, or from a ‘philosophy of mind’ perspective.  The philosophical answer I already know: the car will never be conscious, no matter how many special features it has. There’s simply too many hurdles to take. (Frame problem, symbol grounding problem, qualia problem, Chinese Room problems, Fodorian problems, the list goes on…). So let’s first look at the engineer’s point of view, because it seems a little bit more straightforward

Engineers will ask: what is the car’s performance? Can it do things? More specifically: can it do the *right things*. More specifically: Is it capable of doing whatever it is that you need to be able to do if you want to show that you’re sentient?

Here trouble starts already. If we have a good test measure, then we’re happy. Engineers can design a good test to find out whether the car meets its challenge, provided that the challenge is measurable.

So what do you need to be capable of if you should want to be conscious? We don’t want this question turn into a philosophical one of course, so we need to look at observable measurable behaviors. Do we have examples of conscious systems? What do they do, typically? Well, *we’re* conscious. So what do we do?

More problems. We do SO MANY THINGS. What is the relevant aspect of our behavior? What is the property of our behaviors that signals consciousness? Hard one to answer.

Let’s turn it around. What do systems do that are NOT conscious? Perhaps we can substract all of their regular behaviors from the set of our behaviors and see what’s left.

Rocks. For example. Rocks are not conscious. What do rocks do? Well, they sit still. They wear out and turn into sand. And they respond to gravitational forces in the Newtonian way (throw one and see).

We also sit still. So we shouldn’t take that as a sign of consciousness. And the fact that we turn to sand at some point (ashes to ashes) also should’t be of interest. And our response to gravity: not important. Right: we can cross of at least some from our list. Let’s see what’s left. Anywhere near the crash-test laboratory set-up?

Bummer. Still too many behaviors left potentially relevant for consciousness. And apart from rocks, I don’t know many other examples of systems of which I can safely say  that they are definitely NOT conscious. Insects for example. Do I know for sure they are not conscious? I’ve read papers arguing for the consciousness of E.Coli bacteria. There are actually people (mainly in the 19-seventies though) telling me that the earth itself is somehow sentient, and that’s a rock!

I’m sorry. I cannot give the engineer any good definition of conscious behavior that he could use in a test. We’re thrown back into philosophy, even if we deliberately tried to avoid it.

Reflecting on this exercise I think perhaps the question “Is the Google Car conscious”, should be reinterpreted as aiming for something different altogether. Up until now, we’ve been discussing what would be a reasonable argument for or against the thesis that the car is conscious. Perhaps the whole idea of a reasonable argument is the problem. We’ve just found out that it is very difficult if not impossible to give reasonable arguments, simply because we have no clue what would count as conscious behavior and what not.

But we could also use the ‘case’ of the Google Car a different way. We could ask ourselves, on our gut-feeling: “Right from the hart, is the car conscious or not?”. Personally, I would say no. Perhaps you would say yes. We could either decide democratically (ask 1000 people). Or we could ask the most emininent professor in the room, provided s/he’s able to give us an skilled expert, gut-feeling answer (and not an argument based on reason).

Then, once we’ve decided first whether the car is conscious or not, we now have a different situation and a different engineering question to ask. Suppose, for example,  we decide the car is conscious. We now have a system, completely open to us (since we’ve built it ourselves), and we know it’s conscious. So now the question becomes: what made it conscious? That is an interesting question. And in our attempt at answering it we actually might learn a lot about consciousness.

I think it is the sort of question cognitive science actually has been trying to answer all along, be it about consciousness, or memory, intelligence, emotion, or motor planning. It is a question that stems from creating a working hypothesis about  a mechanic model (this model *has* quality A) and then doing the reverse engineering job of trying to find out what in the mechanics made it such that A is present.

It’s not really about the real thing though. It is a thoroughly pragmatic affair. We’ve first *decided* (based on no rational argument) that the model has A, and only given that hypothesis we analyse the system in the way we do. But I think it is a good way of doing science.

And quite designerly at that! Cowabonga!

Popularity: 17% [?]

Old news, but new to me, a piece of rat-tissue was trained to control a flight simulator!

What I understand of the web-article is that they used biological neurons to create a network of cells in the same manner as they do in software ‘neural network’ systems. Artificial (software) neural networks work on some kind of local learning principle, for example some kind of error-feedback learning. Imagine you give a network some kind of input, and it will generate a (random) output. You then tell the network whether it did right or wrong and the network will adjust the strenghts of the connections between the neurons a bit, where ’succesful’ connections become stronger and the ‘wrong’ connections become less strong. Step by step the network will learn to associate each possible input with the correct output. In cases of a flight simulator the situation is a bit different but the principles stay the same. Here the network works mostly like cybernetic control system where the succesvalue of the output is continuously calculated and continuously fed back into the system as a constraint, driving the system towards the right kind of behavior. Reinforcement learning might be what these researchers have used here but I don’t know about the details of the study.

So now they used biological neurons instead of software neurons to do the same trick. BUT, that of course doesn’t mean that this is the way these neurons work in real brains. The researchers *used* the neurons to create a specific part of a system that is designed by researchers (an artificial neural network). It’s still artificial. For example, the error-feedback procedure used by the researchers might not at all resemble the way real brains correct their activation on basis of error feedback (if they do so at all). The big problem of understanding real brains is: WHO determines what is ‘right’ or ‘wrong’? Who puts the ‘value’ on the feedback signals that are to be used to change the network connections in the network? In other words, how does a real brain know it is ‘in a flight simulation game’ and how does it know ‘what is good and bad’ in a flight simulation game? If nobody tells the brain if the direction it’s going is good or bad, then it cannot learn anything. But we do learn, and nobody ‘tells’ us. In the end it has something to do with survival, with ‘keeping yourself satisfied’. At the moment you do something that gives no satisfaction, then the brain will probably receive this as an error-signal. But then the question becomes: how does the body monitor ’satisfaction’? For food intake or body damage I can image how this would work but for “flight simulation games”????

FURTHER THOUGHT Of course not only our body tells us something about ‘good’ and ‘bad’ but also the physical possibilities for behavior in the real world. If you crash your plane, then obviously you did something wrong. The big thing of being ‘intelligent’ however is that you can learn to fly a plane without crashing even once (since crashing will also end you life so you wouldn’t get to much learning experience out of that anyway). So how can we learn from the actual physical feedback we get from the world (e.g. the way the steering wheel feels in our hands and the way the visual flow changes in our eyes whilst we fly the plane) and then project that sensory feedback onto the succesvalue of a task that we do not get *real* physical feedback from, that is, how do we use the ’safe’ information (visual flow, sense of gravity) and use that to learn how to correct our errors *before* we actually crash the plane? In the end, this will probably mean that representation has to kick in at some point. So for example I see how the horizon tips over to the left, and somehow my brain has to take that feedback and make it mean something else than just ‘the horizon is tipping over to the left’. My brain has to see it tipping and then ‘conclude’ in some way that this *means* that my plane is ‘going to crash soon if I do not act quickly’. Gibsonian, ecological psychology would hold that representation is not necessary here, I don’t know who’s right. Still even in this case I still miss the ‘good’ and ‘bad’ valorisation: if I see the horizon tipping to the left (and feel gravity changing at the same time), how do I know this means I am doing something wrong? I still won’t know until I actually crash, right?

Popularity: 18% [?]

Ik zocht naar een gadget die Sjoerd van Van Berlo mij liet zien, een klein cameraatje achter je oor dat hetzelfde doet als NOOT, maar dan met video. Alleen verkrijgbaar in Taiwan geloof ik. Ben de naam even kwijt. Hoe dan ook, wat ik in plaats daarvan vond was dit . Een filmmaker die in zijn lege oogkas een nep-oog met daarin een camera wil installeren. Weer een small step voor de cyborg, maar een big step voor (het verdwijnen van) mankind.

Er is een gevaar met metaforen. Kees, mijn hoogleraar, hield het mij onlangs nog voor (en ik *wist* het al, kan je nagaan. Ik heb studenten in Den Haag enkele jaren lang vermoeid met het feit dat het oog geen camera is. De mensen die het tentamen gemaakt hebben moeten de vraag nog weten: noem drie verschillen tussen de werking van het oog en de werking van een camera. Zo gevaarlijk zijn ze, die metaforen) Zodra je zegt: een camera werkt zoals een oog, dan ligt de omkering al op de loer: het oog werkt als een camera. En dat is niet zo. Een oog werkt helemaal niet als een camera. Noem de drie verschillen.

Maar de werkelijkheid heeft de filosofie ingehaald. We maken het nu zelf waar. Ons oog wordt een camera. We vermaken (als in, je kleding vermaken) onszelf, tot precies die metaforen aan de hand waarvan we onszelf eerder hadden gedegradeerd tot louter mechanismen. Het oor wordt een microfoon. Het oog wordt een camera. De neus schaffen we gewoon af: wat moet je daar nu nog mee in een informatiemaatschappij? Tast wordt ‘detectie’. Lopen wordt een gekke vorm van rijden zonder wielen. Praten wordt emailen zonder toetsenbord. Het evenwichtsorgaan wordt de accellerometer in onze iPhone. En ons brein wordt vanzelf de database die het niet was, maar waar we dat arme orgaan wel al die tijd mee bleven vergelijken.

Als de tijd gekomen is, dan zal de beoordelingssoftware in mijn brein alle tentamenvragen over het oog alsnog met terugwerkende kracht goed rekenen.

Popularity: 26% [?]

Sophie stuurde me deze

Drawing Robot at the MU, Eindhoven, Netherlands from Nils Völker on Vimeo.

Popularity: 23% [?]

Yesterday I overheard colleagues talking about a robot, I couldn’t find out if it actually exists or whether it was presented as a concept. The one colleague said to the other (if I remember correctly):

“Picture this. The robot is wandering around in the factory. It doesn’t even know where it is, it only knows how to recognize the object and it knows how to act upon it”

(I think the robot he was reffering to is what they call a ‘pick-and-place’ robot that is able to do fast and acurate placing of micro-electronic elements, because that is what these colleagues’ research is about)

“So, this robot is wandering around, not knowing where it is, and at a certain point it encounters the object it must do the placing on. Of course it must first recognize the object as being the object. Then, upon recognizing the object, it will process the visual input and determine it’s own position in relation to the object. It will then recognize that it has not positioned itself correctly in order to be able to do the placement. So, it must know how to reorient itself, carry out that movement, and then place the unit”.

I think Gibsonian perception theory and embodied philosophy could provide a new angle on this story. An embodied philosophy would, I believe, not state the problem as it is framed above. Embodied philosophy would sort of turn the whole story upside down. According to embodied theory, the proces that in biologcal organisms underlies object recognition is not an ‘object recognition module’. Instead, it is precisely the re-orienting proces that ensures that you position your body ‘in the right way’ with respect to the object, that helps you recognize it for what it is. In fact, the gradual repositioning of your body with respect to the body *is*, to all ends and purposes, equivalent to “object recognition”.

Another way of putting this is that ‘recognition’ of an object is first and foremost ‘recognition of the object’s affordances’. The affordance of the object, that is “that set of behaviors that the object directly ‘afford’s, or elicits”, immediately set in motion in the organism reorienting movements that reposition the body towards, what Merleau-Ponty called, ‘maximal grip’. But this is, as I see it, a reciprocal relationship. So just as we can say that the object elicits in the body reorienting movements, we could also say that repositioning of the body is the process the organism uses in order to perceive the affordances in the first place. Both affordance and the behaviors that couple to it iteratively influence one another and just when the organism has taken the right perspective in order to be able to effectively ‘deal with’ the object, it will also recognize the object for what it is. And what it ‘is’, the object, *is* what the organism finds meaningful in doing with it.

So if this robot’s place on earth is to pick-and-place stuff on this object, then, at least according to an embodied philosophy,  should be functioning in such a way that it would *not* first recognize the object for what it is, but that it would use visual input patterns directly in order to continuously reorient itself towards the ideal position for doing the placing movement. And achieving maximal grip (positioning the robot such that it can do the placement) then amounts to “having recognized the object for what it affords”. The whole “object recognition module” in my colleagues scenario can be by-passed. What the robot would need, I speculate, is a layered, behavioral-based architecture similar to the architectures of Rodney Brooks’ robots. One layer above the ‘wandering’ layer would be an ‘orientation’ layer that would try to get maximal grip on objects that afford ‘placing my unit on’. I wonder what my colleagues think about this idea, because it is probably far away removed from the conventional perspective on robot control, in which embodiment and affordances are still relatively unknown concepts.

Popularity: 18% [?]

Well, that is indeed a very intelligent question. Let us discuss that further over drinks, shall we? By the way, do you think I am the *real* Ishiguro or do you think I am the android?

Popularity: 20% [?]

Hey see that intelligent guy asking that intelligent question to Hiroshi Ishiguro (the robot-builder that made a copy of himself), who was giving a key-note talk at CHI-NL (computer-human interaction conference)

Chi-Nederland conference 2009

Popularity: 19% [?]

Ik neem alles terug. Nederland is gewoon de beste club. EK Robotvoetbal!

Popularity: 22% [?]

waar mijn internetprovider zich al niet mee bezig houdt (en hoeveel goedkoper zou mijn abonnement kunnen zijn zonder al deze onzin?)

Popularity: 20% [?]