Sunday 8 June 2014


Macrocognition: Situated versus Distributed

BRYCE HUEBNER
Georgetown University
Department of Philosophy 

VIDEO



OVERVIEW: 'Macrocognition' has two distinct, but closely related meanings. Cacciabue and Hollnagel (1995) introduced it to denote the study of cognition in realistic tasks, where people interact with various forms of environmental and social scaffolding; Klein and colleagues also used it to understand how people manage uncertainty and make sense of real world environments. I introduced a second use (Huebner 2014) as shorthand for system-level cognition implemented by integrated networks of specialized computational mechanisms, whether in individuals or groups. Macrocognition has one sense that's closer to 'situated or extended cognition' and another that's closer to 'distributed or collective cognition' but they are often conflated. There are important differences between the hypothesis of collective cognition (HCC) and the hypothesis of extended cognition (HEC). Recent work on situated and collective memory and philosophical approaches to coordination and planning suggest that HCC is more plausible if we abandon HEC in favor of an 'ontologically thinner' approach to situated cognition. There is a form of collective planning distinct from the planning that relies on web-based technologies and other forms of social scaffolding. Distinguishing two forms of macrocognition, one situated the other distributed, can help us to make sense of a number of theoretically and empirically interesting phenomena.

READINGS:
Huebner, B. (2011). Genuinely collective emotionsEuropean Journal for Philosophy of Science1(1), 89-118.
Huebner, B. (2014). Macrocognition: A Theory of Distributed Minds and Collective Intentionality. Oxford University Press.
Klein, G., Ross, K. G., Moon, B. M., Klein, D. E., Hoffman, R. R., & Hollnagel, E. (2003). MacrocognitionIntelligent Systems, IEEE, 18(3), 81-85.

46 comments:

  1. You might want to watch out if you are integrating 14 cups of coffee a day with your "cognitive self".

    See this post (a link to the primary source can be found by clicking(:
    http://usefulscience.org/post/high-coffee-consumption-correlated-mortality

    ReplyDelete
    Replies
    1. I've heard that Voltaire drank as many as 100 cups per day!

      Delete
  2. You used Shakespearean actors as examples for how we used social and environmental scaffolding as triggers without remembering details. I find this problematic. For one, many of Shakespeare's plays are in a particular meter, which means that each word has been chosen specifically for its cadence and order. The picture you chose was from Midsummer Night's Dream, which is noted for Shakespeare's use of trochaic tetrameter. Even one mistake in which word the actor uses will ruin the desired effect. In this case, actors must remember every detail, so I believe this goes against your idea of using triggers to remember the gist.
    Additionally, many of shakespeare's characters give soliloquies, either as inner reflections or asides to the audience. In the case of inner reflection, the actor has no social scaffolding-- he is reciting hundreds of lines without interacting with any other actors. Even when saying lines to the audience, I would expect that the audience changes every night so I'm not sure how the actor could use an an audience member as a sort of trigger for remembering his lines.

    ReplyDelete
    Replies
    1. Nonetheless, the actor knows when to start the monologue based on the previous scene and remembers the tone which follows another actor's tone.

      Delete
    2. There's actually some scholarly dispute over the extent to which actors in the globe were fully scripted. I'm not a specialist, but I gather that some people have argued that the written versions were written separately from what the actors were acting.

      Delete
  3. Dear Bryce, Thank you very much for your philosophical presentation! As a prior student in philosophy, I felt some rich and deep stems from traditional “meta-physics” preoccupations for the question of “onto-logy” into your “macro-cognition” research. Could you give the name of the three philosophers who influenced you the most? Don’t you think the object of “meta-cognition” studies is a materialistic and individualistic version of collective control ambitions of “meta-physics” and associated epistemologies?

    ReplyDelete
    Replies
    1. Spinoza, Darwin, and Dennett. And yes, I see this as a naturalistic and scientifically inspired alternative to traditional forms of metaphysics and epistemology.

      Delete
  4. What is the web Good For !

    Can we imagine new generation of decision-making tools that can use the web us a big knowledge storage. These tools can update their decision tree by seeking, bringing and incorporate new decision rules from the web!!

    ReplyDelete
    Replies
    1. I'd love to see something like that, and think more about it! Are there things in that vicinity that interest you?

      Delete
    2. Yes, I ‘m doing work on this topic by creating decision-making system based new knowledge organization. This system attend to be able to integrate new knowledge from outside the system.
      (Application to support the transfusion threshold)

      Delete
    3. Very cool! I'd love to see it!

      Delete
    4. I'm in my first steps and I will let you know about my progress.

      Delete
  5. 1. Does your understanding of participatory metacognition supports the theory of enactive cognition? If not what are the differences?

    2. In our work at the Global Brain Institute , we are investigating the concept of stigmergy that is basically using the agent's environment to encode relevant information by directly interacting with the environment. For example: the way ants collectively form trails using pheromone traces on the ground. Would it be possible to understand the web as a stigmergic space that help human communities to coordinate collective metacognitive activities?

    ReplyDelete
    Replies
    1. I think that project has a lot of promise, but I think it's a nice way of thinking about cognition individualistically. In fact, I start my book with a discussion of that sort of case, and explain why I don't think it's a case if macrocognition. As for enactivism, I think that it's a mistake to eliminate talk of representations from the cognitive sciences. I have a lot of sympathy for some of the aspirations of the enactivists, but in general I'm skeptical of their methods.

      Delete
  6. Very interesting talk. I strongly agree with the idea that situated and collective forms of distributed cognition are often wrongly conflated and should be distinguished. My question for Professor Huebner is whether the conception of distributed cognition he defended in his talk argues against the idea that the Web constitutes a kind of “Global Mind.” Does the embedding of cognitive agents in a shared collective information processing structure entail that the whole distributed system constitutes a “higher-level” mind?

    ReplyDelete
    Replies
    1. There may be a global mind, I don't rule that possibility out, but I don't know the arguments in support of that claim. Unfortunately I missed the talk on the global brain, but I look forward to watching it soon!

      Delete
  7. It was a very interesting talk. I would like to know more about social cognition and metacognition, could anyone give my some references please?

    Also I'm wondering if we could say that memory capacity depend on some habits?

    ReplyDelete
    Replies
    1. Check out:
      Gallotti & Frith, http://www.sciencedirect.com/science/article/pii/S1364661313000417

      And,
      Shea et al, http://www.ncbi.nlm.nih.gov/m/pubmed/24582436/

      Delete
  8. Very interesting talk.
    At the beginning of the talk the speaker said that it is difficult to imagine accessing information somewhere in Helsinki as 'extending the mind' through the mobile phone. But the biggest part of the talk was dedicated to grounding the statement that memory function is dependent on associative relations with immediate environment of an agent that memorizes (Shakespeare's play). Mobile phone just extends the situational context that is used for reconstructing memories. The question then is what is the problem of imagining this as 'extended mind'?

    ReplyDelete
    Replies
    1. I think the problem with imagining the server in Helsinki or cellphones as part of the mind, is that we open the doors for everything in our environment to become part of our minds, whether it's the coffee we're drinking, our cell phones or the chair we're sitting on, and the term "mind" starts to become meaningless. What Huebner seems to suggest is that we have to view our minds as being situated in an environment, meaning that our minds never operate independently of an environment anyway, and this is why he thinks we should adopt a participatory framework.

      Delete
    2. Thanks for your answers/comments.

      I do not think that being more connected and being situated are mutually exclusive.

      Our sensory organs receive much more information than can be processed or attended even without mobile phones, therefore an important cognitive function is selection for relevant information (selection for relevance, or information filtering). Getting one more sensory modality does not mean that a person suddenly stops selecting relevant signals which seems to be hidden under the statement that ' the term "mind" becomes meaningless'.

      Delete
    3. Agreed! And I also think our technologies are often developed as strategies for pre-filtering sensory information!

      Delete
  9. I would guess that while there are similarities between the transactive memories of long term couples and the extended memory using Google, the fundamental motivations for why the memories are linked is different. I would offload my photos and memories to Siri in order to access it later, whereas I would tell my friend or lover my memories because we have a human-to-human relationship. It is an added bonus that my friend might remember my memories for later, but I don't think I would view my extended memory in the same way I would rely on Google.
    Do you view human-to-human macrocognition in the same way that you view human-to-technology macrocognition?

    ReplyDelete
    Replies
    1. I don't think that I believe there are many cases of human-technology macrocognition. There may be some cases, but I've not yet been convinced. I'm bothered by the assymmetric flow of information in these systems (NB: there are plenty of cases where other people are used as tools, and I don't see those as cases of macrocognition either).

      Delete
  10. You mentioned abandoning the observer perspective and considering cognition from the individual perspective, claiming this should generalize even to eukaryotes, in the pursuit of good cognitive science and good biology. That sounds a lot like the autopoietic view. How do you think such an approach to biology or cognitive science fares against Stevan Harnad's criticism of it being nothing more than "biological hermeneutics".

    See the following:
    Maturana's Autopoietic Hermeneutics Versus Turing's Causal Methodology for Explaining Cognition
    http://eprints.soton.ac.uk/263810/3/kravrep.pdf

    ReplyDelete
  11. Dr. Huebner’s writes “I argue that there is substantial philosophical and empirical support for the existence of collective emotions” in the abstract of his paper Genuinely collective emotions. To discuss such a topic we should talk about what an emotion is.

    I believe an emotion is something that is both felt and observable. For example, if John is happy, not only does John feel happy, but someone else could also assume that John feels happy based on his observable behaviour.

    However, if John is happy, but I think John is sad based on my observation, John is NOT sad all of a sudden. What I perceive John’s emotions to be does not make John feel them.

    Moreover, John may look sleepy when he’s knocked out with deep anesthetics. However, John is not sleepy as he cannot feel while under anesthetics (probably).

    Extending this further, if Germany as a whole cannot feel, or be conscious (which Dr. Huebner agreed with this principle during the discussion after his talk), than how can Germany be happy upon winning the world cup. Yes, to observers Germany would look happy, but as Ishan points out in an earlier post, this is hermeneutics. We ascribe emotions to many things in our environment. However, if an entity cannot feel (i.e., be conscious, have intentionality, have a mind, or any other synonym), how can it have an emotion.

    ReplyDelete
  12. I think that all of the capacities that are relevant for the production of an emotion can be distributed. I've never been convinced of the claim that an emotion must be felt to generate the right sorts of computational and behavioral states. To be quite honest, my guess is that most emotions are non-conscious. I agree that the commonsense category 'emotion' has a 'feely' component. But I think doing good science on emotion requires abandoning that component, and adopting a view that focuses on action plans, affects on cognitive processing, and the production of behavior.

    As for Germany, I think that many Germans and Mannschaft-sympathizers were happy, and many Albiceleste fans were sad—my self included. But neither Germany nor Argentina were in emotional states.

    ReplyDelete
    Replies
    1. I agree that the common sense category of emotion has a ‘feely’ component. Beyond that, I believe the common sense definition of emotion requires feeling. The definition of emotion according to Merriam-Webster, Oxford, and Robert-Collins dictionaries all include ‘feel’ or ‘conscious’. While standard dictionaries are far from ideal for academic debates, they inform us about how non-specialists will interpret our writing.

      If most emotions are non-conscious, and we should ignore feeling to do good science, we should abandon the word emotion in this research. This is straying far enough from the common-sense definition of emotion to confuse both researchers and non-scientist. Emotion in this sense is nothing more than a best-guess based on observable behaviour. If I think you are happy based on your actions, but you truly feel sad, what is your emotional state? If we drop feeling from emotions, than I am correct; you are happy.

      Delete
    2. I think we're just going to disagree here. My guess is that there will never be a case like the one you describe, once you take into account action plans, affects on cognitive processing, *and* behavior.

      Delete
    3. The case I describe will occur inevitably in real life. If we ask 100 people how John feels, someone is sure to say that John has an emotion which John does not report having.

      Perhaps we could guess John's emotional state with a high certainty if we were measuring many physiological ongoing from John, knew his life history, how he acted in previous situation, and what his intentions were. Even if we could predict emotions with perfect accuracy after acquiring all these details, this scenario is improbable.

      Is it possible to take into account all John's action plans, cognitive processes, and behaviour? Is it possible to do the same for a macro scale system? If not, we will inevitable ascribe emotions to these entities (e.g., John or Germany) which are absent in the entity.

      Delete
    4. I'm not particularly concerned with non-expert opinion here. You're probably right that some people would have a hunch about the person's emotional state that differed from the target's actual mental state. But, at least in part, that would be because these people were making their best guess in light of some subset of the available information. Emotions are psychological kinds, and we do pretty well figuring out what emotional state a system is in familiar cases. But as we move further and further away from what is most familiar, we need to rely on psychological data, rather than hunches. Hopefully that helps to clarify my view.

      Delete
  13. Dr. Huebner: what exactly do you mean when you use the term "narrative structure"? I have a general idea of what this can mean, but I've never heard it used in this context before, so I just want to make sure it's clear to me!

    ReplyDelete
    Replies
    1. That's a hard question, and a version of the one that Richard Menary asked me. Basically I'm just thinking of representations with an agent-event structure, typically encoded linguistically. Menary wanted me to clarify this, and I guess this may a better way to do it. Many skeletal representations are encoded associatively, as predictions regarding state-value relations or action-outcome contingencies. We flesh these out using systems that construct models that represent ways the world could be. Those representations are the ones that "narrative structure" was supposed to stand in for.

      Delete
  14. There's individual cognition, and then cognition performed by an individual organism. In the first case, you get an individual which has integrated beliefs and desire, and thus a sort of unity in its goals and motivations, and which will address challenges accordingly. The organism, however, is a subpersonal unity. It has no intrinsic teleology (which belongs to the personal level). It does cognitive work, some of which participates in what the individual, some of which doesn't.

    I think the danger with merely embedded or situated cognition is that it may conflate these two. Even though your (very important) distinction between individual cognition and macrocognition may account for the fact that our behaviour is often to be interpreted in terms of group dynamics rather than individual cognitive dynamics, it doesn't entail that we can suddenly infer things about the cognitive system at the subpersonnal level of description from things we know at the personal level. I.e. we cannot infer that the realizer of the individual cognizer is an individual organism.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. I'm sorry, I'm not expressing myself clearly at all. I think the distinction between individual cognition and macrocognition undercuts an argument I could make against merely situated cognition, namely that it can't account for group cognitive dynamics. But it doesn't undercut another argument, which is that you cannot count on isomorphy between personal and subpersonal level.

      Delete
    3. I think that we are in agreement. I think you need to start from a task analysis, which may occur at the level of collective or individual behavior; only once this is in place can you ask: what is the integrated network of information-processing systems that is responsible for success at that task. But in every case, I think those stories will make reference to the individuals involved, either as components of a larger system, or as exploiters of locally salient information.

      Delete
  15. This comment has been removed by the author.

    ReplyDelete
  16. This comment has been removed by the author.

    ReplyDelete
  17. One more question :) - I’ve read some of these ideas before in “I am a strange loop”, by Hofstadter - I think you were both influenced by Dennett? In particular, Hofstadter talks a lot about how raising a child can be a unifying goal which helps to further align the behaviours and beliefs of a long-term couple, which really makes sense in the context of everything you talked about today. He also talks about how these connections between people are low-bandwidth, and low fidelity, which could prevent the elaboration of a conscious entity. Anyway, I was just interested in if you have any thoughts about Hofstadter’s strange loop with regards to consciousness, and if it has any applicability to the topics you discussed today.

    ReplyDelete
    Replies
    1. I don't think a lot about consciousness, that's the one thing that Dennett and I have never really talked about. But I do LOVE Hofstadter's work, and I'm sure that it has really inspired me—even if nonconsciously

      Delete
  18. I asked this in the review session, but I'll post it here (a) because Professor Harnad said we should and (b) I'm curious if you have new insights about it:

    So far, we've agreed that human-human macrocognition can exist, but probably not human-technology macrocognition. What about inter-species macrocognition? The example you mentioned was the symbiotic relationship of the bobtail squid and bioluminescent bacteria. To what extent are they cognizing together?

    ReplyDelete
  19. If the bobtail squid and bio-luminescent bacteria are cognizing together, I think we have to extend this label to many other symbiotic relationships. Are the bacteria in my gut cognizing with me? If so, are they part of my consciousness, or does our partnership form another mind with distinct feelings?

    ReplyDelete
  20. I wonder what benefit there is to distinguishing between collective and extended cognition. While I understand the differences between the two that you are proposing, for me, it seems a lot less complicated to simply say that whether it is just me thinking, or interacting with another human being in a collaborative manner, or using the computer to find information, myself, the other person, and the computer are all part of a cognitive system that is created for the purpose of some goal. Granted, your notion of collective cognition is a little less egocentric than a purely extended view.

    ReplyDelete