Dennett sees intentional ascription as a convinient pragmatic solution for making sense of our environment, but takes such interpretation to be merely a less precise and detailed version of any other "stance" (as I understand it). A physical, design or intentional stance is all about the same, and differ only with respect of accuracy and computational strain. A system, such as a rock, has behavior simple enough that the loss of information under an intentional interpretation is less warranted than a physical interpretation, and vice versa for computers, dogs and humans.
Is there really nothing else to intentional interpretation (putting other issues aside) than prediction? It seems to me that there are some crucial differences:
(1) Often more than one interpretation will have the same predictive consequences.
(2) Often we don't interpret to predict.
(3) Often intentional interpretation opens up more content than other interpretations.
For instance, two predictively equivalent interpretations may result in cognitive access to vastly different scenarios. I may rationalize a systems behavior under ascriptions that makes it either rational or irrational, and under one interpretation we would say that the system is competent, worth relying on etc., the other would make it incompetent and unreliable. The difference need not be a matter of predictive succes, yet it makes a huge difference. How can we account for that without going further than Dennett's deflationary account?
We can for instance interpret a sports player as faking poor performance - perhaps (s)he is trying to honour an agreement to let in a lot of goals in exchange for a payment from a mafioso. We could say that the player is incompetent, or is competent but trying to lose. Here the predictions would be the same (roughly), but the scenarios are incompatible and different. One is true at the expense of the other.
When it comes to industrial robots, and if we only want to understand the systems intrinsic states, then it seems like we lose nothing under a mechanical interpretation. But we lose something under such an interpretation of social or psychological behavior. The robot does X under circumstances Y, and we need nothing more to handle the robot. But that the player does X under Y may mean that (s)he is incompetent or illoyal. This has very different consequences on appropriate interaction with the system.
What are the functions of intentional interpretation of systems? Can Dennett's account really explain it all, or is there more to intentionality?
Hello Luca,
what a contribution. Worthwhile thinking about it, especially now that we have the scientific and industrial evolution, where human-robot interaction becomes a relevant question.
Your introduction reminded me to Aristotle's philosophical description regarding the meaning of things, and it's rejoinders in political ideoligy (e.g. Feuerbach). In their theories the essence of a being is its' presence and function itself. 'You are as you will be', as Nietzsche said. Their essays kind of point on a discussion on whether 'reality' is mere action, regardless their intentions. In philosophy intention is rooted in a different domain (Frege) who is also a mathamatical philosopher. Meaning, different from being, was always an issue. Finding truth in complex systems only out of logic represents the opposing pole in this debate. I am not aware of any philosopher who was able to manage an integration of both. Of course, this relates to Descartes' 'brain in the tank' and the outer world problem, i.e. how can we actually know that every thought, and thus intention, is an illusion. His answer: Cogito ergo sum, does not answer the question, whether it really matters. Anyway, all these philosophers may provide (more or less technical) frameworks on how to deal with rational or irrational intentions/actions, and the (im)possibility of consciously separating them. In fact, one has to decide on a perspective or creating a new one, and I think you are doing well with it :)
Back on the topic:
Here is a counter-question, assume that a human or a robot is showing malicous behavior. Can we predict this? Well, when we know how the cognitive systems work, yes. Now, if we could explain it then making predictions from it, what then? Is this a matter of pre- or post-action (justice taking)? Or in other words: Does knowing the difference change the things? Only when we intervene. Why should we know about the cognitive systems, if not for prevention. That is first and importantly true for robots. Where any kind of malfunction is based on system errors or malware. The latter could be true for humans too, but a similar conclusion (juridical prevention) entails the idea of guardian- or trusteeship. For this kind of intervention into human life there have to be at least political contracts.
And this brings us back to Feuerbach and Marx (keyword: historical materialism). There are two accounts on human-system interaction: 1. Top down: We shape our environment. 2, Bottom up: Environment shapes us. Simple conclusion. We shape ourselves. Now this is an optimistic interpretation: Marx wants to have the best development for every human being (education and quality life-time). 'Intention' as stemming from 'insight into the the necessity' is a central metaphor in the political advance of this idea. The logic nevertheless can be applied to every political care taking system. Now are there some kind of guidelines on how we treat robot or computer systems? Are there system definitions in human society which may serve as blue-print? Yes of course, every law, defines how to treat system mal-functions. Can this also be used in describing positive behavior. This is what I would also ask for. I think there is no blue-print for being positive, for example, if your basketplayer shows high level performance, would you ask why? Still yes, when we are not just focused on outcomes (which is actually not true in every human...). Before I go on, let me say that I also think that there is more about understanding human systems, but for prediction. However, I think, the analogy with robots works out well, when the focus lies on shaping the system-agent interaction in an intentional way, meaning, we can not prevent system-errors by understanding their causes. This is simply unethical (from my point of view). But we can install systems (societal and computational) which acts reasonable with their respective agents... oh ups... for a minute I thought of central networt-robot interaction instead of human society-robot interaction... scary.
Best wishes,
René
Luca,
Thank you for your description of Dennett's intentional stance.
A few thoughts first from the viewpoint of neuroscience. Our minds are associative patterners, always complexing information (consciously and unconsciously) for the situation at hand in anticipation of the outcome of our actions, that is, creating knowledge (the capacity, potential or actual, to take effective action). While this is perhaps not exactly "prediction" in terms of exactness in time, nature and magnitude of future events, it is certainly predictive in nature.
Now how we define intention comes into play here. If we see intention as the source with which we are doing something, the act or instance of mentally and emotionally setting a specific course of action or result, a determination to act in some specific way (deliberate intention), then it is a choice of the self relating to the world.
Searle (1983) believes that people have mental states, some conscious and some unconscious, which are intrinsically intentional. These are subjective states that are biologically based, both cause by the operations of the brain and realized in the structure of the brain. Thus intentionality is not a description of action but is in the structure of action. Thus if you have an intention, it is an intention to do something. However, states such as those represented by beliefs, fears, hopes and desires insinuate intention, and they are about something. The ideomotor model of human actions contends that human intentions are the starting point of the actions associated with those intentions (Prinz, 2005). Conversely, the sensory-motor model of human actions identifies sensory stimulation as the origination of actions.
Does this insinuate a causal relationship between the external event and the desire? Searle contends that causality is generally considered a natural relationship between world events, while intentionality is not a natural phenomenon, not part of the natural world, rather something transcendental, something that stands over or beyond, far beyond objectives, plans or strategies, even beyond hopes and expectations.
We take exception in the relationship between intention and expectation, with expectation representing the juncture of where we currently are and where we want/desire to be. While intention includes the purpose and attitude toward the effect of one's action, the outcome--with purpose implying having a goal in mind or the determination to achieve something and attitude encompassing loyalty and dedication to achieving that goal--living (thinking and feeling) is not about the outcome. Rather, the focus is the continuous creation process of the relationship between intention and expectation. (As continuous learners, we remind ourselves that we are verbs, not nouns. Our early learning that a noun is a person, place or thing needs to be tossed aside!!!) Knowledge is always partial or incomplete, context sensitive and situation dependent, and, as we choose, we are in a continuous spiral of expansion. Intention serves as a tool for that expansion.
[These are excerpts from an upcoming book on change that will be made available to all researchers on this platform in about 6-8 weeks.]
Thank you all for the good answers. I will reply each of you when I have time. I also wish to elaborate a bit on my points and the implicit argument against an overly performative or instrumental conception of intentionality. I have no positive account in mind, though, so this is mainly an attempt to air some of the qualms I have with certain more or less popular views in currency at the moment. But again, thank you for your answers, all of them are interesting!
And as for the last reply by Alex Bennett: It is nice that you bring up Searle. I personally like his theories and writings, but I am not really convinced by his theory of intentionality. But in my mind it is more correct than Dennett's, though Dennett correctly identifies at least one of the important functions of our representations of intentional features in our environment. And I agree that in a broader sense, intentional ascription may be described as predictive. But I will try to make my point (and question) clearer, since one can relax the (perhaps) too strong notion of prediction I had in mind and (if I am not mistaken) rephrase the question/challenge in terms of that weaker notion.
I will return with a better reply one of these days.
Kind regards,
Luca
PS. Thanks Susan Laird for the link, that is one of the texts where Dennett's position is presented and which I had in mind.
It has been a while since anyone attended to this question, but it is a very good one. Maybe someone will appreciate my response.
In order to understand the future, we must understand how things work. Rust, for example. We see from experience that oxygen intends to join with iron. We can describe the process by which this happens. Conditions of the iron or the oxygen environment tend to increase or decrease this process.
As a result or our observations, and the descriptions thereof, we can predict rust! The logical corollary of prediction is that we can control rust!
This is all based on intention observed in the interaction of oxygen and iron.
Our ability to control anything depends on recognizing intent, or purpose, in the interaction of certain essential elements. Logically understanding those interactions creates our ability to predict their behavior and control them.
All based on a logical understanding of conceived, hypothetical, intent or purpose.