Sunday 8 June 2014

Transactive Memory and Distributed Cognitive Ecologies

JOHN SUTTON
Macquarie University
Department of Cognitive Science

VIDEO



OVERVIEW: Does the internet alter the way we remember? What understanding of memory makes sense in light of our rich interactions with technologies and with other people? This presentation introduces theoretical and empirical work on distributed cognitive ecologies as a framework for addressing web science and the mind. It surveys recent accounts of the effect of new technologies on human memory, with a focus on transactive memory theory. It embeds recent empirical findings on the ways we remember in conjunction with each other and with online systems in a broader picture of socially distributed remembering. In place of metaphysical concerns about extended cognition and popular worries about the erosion of natural memory, it suggests a number of rich research possibilities for integrating the cognitive and social sciences.

READINGS:
    Michaelian, K., & Sutton, J. (2013). Distributed cognition and memory research: History and current directionsReview of Philosophy and Psychology4(1), 1-24.
    Sutton, J., Harris, C. B., Keil, P. G., & Barnier, A. J. (2010). 
The psychology of memory, extended cognition, and socially distributed rememberingPhenomenology and the cognitive sciences9(4), 521-560.

http://www.johnsutton.net/Sutton_CHSC.pdf
http://www.johnsutton.net/PCS_Sutton_Harris_Keil_Barnier.pdf 
http://www.wjh.harvard.edu/~wegner/pdfs/science.1207745.full.pdf

35 comments:

  1. 1. "seek not criteria but dimensions" could you elaborate on this? Which dimensions are relevant? does this actually support context dependent kind of approach.

    2. "Many cognitive interactions... involve relative low intensity and rate of interaction" In order for cognition to serve any function towards increased fitness, it seems the environment needs to be relatively stable and unchanging compared to the speed of processing that underlies cognition. In this sense, low intensity of interaction and asymmetry might be necessary to realize web extended cognition. This is not substantially different from how we interact with physical environments.

    ReplyDelete
    Replies
    1. thanks for the comments - just checking I can log in ok to comment

      Delete
    2. Ok great. The search for criteria springs from a quest for necessary and sufficient conditions for (extended) cognition, or for its essence. I don't think we need to identify its essence to study clear cases effectively: the multidimensional approach, yes, is meant to allow us to locate continuous variation of many kinds across different cases - so yes, exactly, highly context-dependent.

      Delete
    3. Very good question about stable environments. What I wonder is whether in special cases - as we gain expertise and fluency with specific cognitive technologies, eg specific regions of web space - we can cope effectively (and flourish in using) even more dynamic and interactive resources - what do you think?

      Delete
    4. I agree. Increased competence is almost always reflected in the speed of interactions and "bandwidth" of information exchange. Still, I think that from a purely computational approach effective cognition requires that the environment or at least the significant aspects of the environment will change slower than the computational processes involved in cognition. If I perceive motion (say a Tiger is attacking me) and cannot respond within a relevant time interval, I will not survive so the very function of perception will not give me any advantage. But if I get it fast enough, at least in some case I will survive.

      Delete
  2. Small question: could you clarify the difference between collaborative and nominal groups? I don't quite understand "pooled sum of nonredundant items."

    ReplyDelete
    Replies
    1. Yup, good question, sorry this went v fast. Say we are remembering word lists. The collaborating group of 3 people may remember 7 out of 10 - jewel, cat, igloo, gem, palace, lion, triangle. The nominal group is 3 separate individuals working independently. Say person A remembers 5 (jewel, gem, diamond, palace, triangle), person B remembers 4 (igloo, cat, lion, triangle), and person C remembers 6 (jewel, leopard, lion, palace, gem, cat). To find nominal group performance, we count up or 'pool' the nonredundant items (ie discount ones that more than one member recalled). This gives us jewel, gem, diamond, palace, triangle, igloo, cat, lion, leopard, ie 9 items: so this is collaborative inhibition, because the collaborating group recalled 7 and the nominal group 9 (even though no single member of the collaborating group recalled more than 6). Nominal group performance is a mere aggregation, with no emergence.

      Delete
  3. Interesting ideas in this talk! Professor Sutton argued that we should abandon the paradigm of “extended” cognition in favor of “distributed” cognition. My question is whether we should still be talking about minds here (as opposed to cognition), when we are talking about distribution and extension. Are minds distributed on Sutton’s account, or is it cognition that is distributed, without the mind itself extending?

    ReplyDelete
    Replies
    1. Good question, thanks, I wasn't clear on this in the talk. I don't like the term 'mind' any more than 'cognition': both are umbrella terms which can make us think we know what we're talking about when we don't. Some treat minds as essentially conscious, where cognition is not: I don't feel the pull of that intuition. So, ideally I would talk more seriously only about specific capacities such as remembering, planning, decision-making, grieving, imagining, perceiving etc (which we have a pretty good grip on in common sense and in theory). In practice of course I talk about 'mind' and 'cognition' all the time, but I think these are imprecise terms: as I said, I don't like the common assumption that cognition excludes emotion, which has historical justification but doesn't fit the modern cognitive sciences.

      Delete
  4. How would you explain "shared encoding" as an external resource in couples? I can't seem to figure it out.

    ReplyDelete
    Replies
    1. Good question. Shared encoding can take a few forms. One is simply going through shared experiences together: in stronger cases, we engage in joint action with shared goals, or in weaker cases we simply happen to witness the same events (like a car accident). Or, shared encoding can be the active joint generation of ideas or material (eg brainstorming or planning together). This is a feature of some kinds of rich interaction in which each person may act as an external resource for the other.

      Delete
  5. Could you explain more how you see the context and multiple context? Thank you JOHN SUTTON.

    ReplyDelete
    Replies
    1. The extent to which remembering (for example) is context-dependent is itself context-dependent. Some times some people shield themselves more from their contexts and rely more on internal resources alone.

      The idea of 'cognitive ecologies' is meant to point us to the complexity of the multiple contexts in which remembering (for example) occurs.

      Delete
  6. Dear John, Thank you very much to share with us a lot of challenging questions. Cognitive ecology suppose a complex net of dimensions. According to computational limits of algorithms and machines, which would your main and possible computationals dimensions urgently to integrate in the context of cognitive computing applications such assisted reading softwares, where memory capacities have an important role?

    ReplyDelete
    Replies
    1. Thanks Ludovic. I don't have a general answer, it depends on the project aims. For assisted reading softwares - which I don't know much about - presumably ease ('transparency') of use is a vital dimension, so that users don't have to worry about the interface all the time but gain (with expertise) proceduralized skills so they can focus on the content not the system. Mutual responsiveness (what Harry called co-adaption) will also matter here.

      Delete
  7. Lorsque vous évaluez les technologies de mémoire pour les vieux couples, est-ce que vous incluez dans le calcul l'affinité entre les deux? Si les deux ne s'entendent pas, alors ils auront du mal à s'entraider en mémoire... surtout s'ils ne se parlent pratiquement pas.

    Translation:
    When you evaluate results of the memory of old couples, do you include couples that have no affinity? What I mean is, I have often seen couples that barely talk to each other, they each live their lives independently of each other. How is this evaluated?

    ReplyDelete
    Replies
    1. Great question Konstantinos. Yes, relationships take many forms and as you say some couples don't have such constantly interactive practices of remembering together. In those cases, standard collaborative inhibition effects occur, ie they perform better in many memory tasks when working alone than together. We just don't know yet what else goes with such cognitive autonomy, for some couples it may be effective: and we predict that it will be easier in such cases when one partner survives the other.

      Delete
  8. You talked about individuals who sort of shield themselves from their cognitive ecologies and try to rely on the cognition inside their heads and others who heavily rely on external resources. Have you done more research on this? I find that spectrum very interesting and I would be interested to see if there were a way to measure that willingness to depend empirically and also look at potential correlates in lifestyle, personality, temperament, IQ, etc.

    ReplyDelete
    Replies
    1. I'm also interested in this question. Undoubtedly, there could be evolutionary advantages to both ways of being, but I wonder what the influence of the environment is in understanding why someone would fall on one end of the spectrum vs. another.

      Delete
    2. Thanks Rachel and Nicole, yes I too am fascinated by this, and no we haven't got very far with it empirically yet. The complication is that this is not only an individual difference variable, but changes across tasks too; and across relationships (some people are very open to or interdependent with their partner but not at all in other social contexts). Addressing such questions requires richer mixed methods, and requires us to expand the unit of analysis beyond the individual (even when sometimes we end up finding that individuals maintain more cognitive autonomy). Keen to hear your ideas.

      Delete
  9. You made a paper a while ago that meant to make a framework for enlarging memory studies, if you will. Now, people in sociology and neurology have very different ontological and epistemological commitments – not mentionning their unwillingness to talk to each other. How could we get these people to speak together?

    ReplyDelete
    Replies
    1. Thanks Louis, yes it's important not to underestimate the difficulty of working across concepts, methods, and tacit assumptions in radically different disciplines. My hope is that using more specific, relatively constrained topics - such as episodic memory - one can pool the resources of the various disciplines in shared projects. In fact I think memory studies is going ok with regard to these interactive discussions - perhaps emotion theory is further ahead, for example, though. In sociology there are enough people interested in cognitive theory, and in neuroscience enough keen to expand the unit of analysis beyond the brain.

      Delete
  10. 1. My approach is that all cognitive phenomena are 'distributed' or 'collective'. Therefore although usage of these terms is probably correct, it is also misleading in sense it draws an artificial boundary between 'extended' / 'non-extended' and 'distributed' / 'non-distributed'. I agree that thinking in terms of continuum between these dichotomies is more useful than trying to classify things according to them.

    2. Related idea is that maybe studying development and evolution of cognitive systems along these dimensions is more informative than thinking in terms of static properties of these systems.

    3. Just an association with a 'Otto with Alzheimer's' thought experiment which seems to be popular today: there is an iPhone app which helps blind people to 'understand'/'see' pictures by allowing to post them to a crowdsourcing site and ask other peoples' help. It looks to me as a true 'extended cognition'.

    ReplyDelete
    Replies
    1. I agree with your third point. According to me since human are using objects we could speak about a kind of extended mind. But of course we have to put some conditions.

      Delete
    2. Thanks for these good points. 1. Yes, exactly, the point of my argument for a dimensional account of distributed cognition (rather than a search for criteria) is to embrace and make progress with understanding a set of continuously varying cases, rather than a clear line. I've argued strongly for this multidimensional framework as more fruitful. 2. Yes, once we have a multidimensional space for analysis and study we can trace the trajectory of particular systems through that space over time. 3. Cool - how is the information fed back to the users?

      Delete
  11. Do you think that it will be a viable option to use the concept of distributed cognitive ecology to design external systems that compensate the memory deficits of patients with dementia as the illness progresses? How could we manage to apply some of the person biographic and semantic memory mechanisms into an external device so that it will keep working in a similar way to the individual's previously untouched memory?

    ReplyDelete
    Replies
    1. Yes, this is the dream or ultimate ideal of the kind of work we're doing now. There is some good research already just trying to describe and find out how older adults already use such systems (eg Wu et al, 'Collaborating to Remember', 2008). I wouldn't say the individual's memory (eg before any memory decline with age) was 'previously untouched', because I think we are always richly cognitively and mnemonically interdependent. The supporting devices then don't necessarily have to work in the same way as biological memory to be effecitve. Thanks for the questions.

      Delete
  12. A very good talk
    I have two questions:
    • Criteria for extended mind are still persisting even we change our point a view ( Dimension) from extended to distributed cognition ?
    • The power of the transitive memory is well shown in your talk, especially in collective resolving problem and remembering, can we see the interaction between distributed cognition and individual cognition as recursive effect influence each one the other ?

    ReplyDelete
    Replies
    1. Hi Eltaani and thanks for the comments. Yes, my suggestion that we use 'distributed' rather than 'extended' cognition is just a terminological recommendation: it's not suggesting a major shift of view or a deep distinction between two types of cognition. 'Extended' has the misleading implication that there's a primary fully internal kind of cognition, whereas 'distributed' clearly suggests a thoroughly integrated heterogeneous system. Yes, individuals are influenced by the distributed systems in which they participate (eg people who collaborated on a memory task do better when later recalling the same material individually than those who never collaborated). But individual cognition and distributed cognition aren't two separate processes.

      Delete
  13. Very interesting talk. I was pleased that Dr. Sutton addressed the confusion surrounding the definition of ‘extended’ and ‘cognition’ and suggested we use ‘distributed-cognition’ as an umbrella term encompassing distributed- remembering, thinking, decision-making, interacting, and feeling.

    We have talked little up to now in this conference about consciousness or feeling. However, I believe feeling is a special case and cannot be distributed. Sure I can have the same genre of feeling (e.g., happiness or touch sensation) as someone else, but we are not experiencing the same instance of feeling—feeling is inherently the property of a single feeling entity (a feeler). If a feeling is distributed between myself and someone else, who is the feeler? If we both feel similar things, this is still two instances of feeling. Alternatively, does this distributed-feeling create a new feeler composed of both of us?

    On the other hand, distributed-remembering makes sense. Two people can remember different elements of an event and talk about it to remember the name of the event. The final product is unitary, the process is shared (distributed). A similar argument follows for distributed- decision-making, thinking and interacting.

    ReplyDelete
    Replies
    1. Thanks Robert, and thanks for the comments. That's exactly right about the unitary project of the distributed process. I haven't worked much on distributed feeling, but there are some promising lines of thought: I recommend papers in *Philosophical Psychology* by Jennie Greenwood ('Contingent transcranialism and deep functional cognitive integration') and by Achim Stephan et al ('Emotions beyond brain and body'); and also Joel Krueger's paper 'Varieties of extended emotions’, in *Phenomenology and the Cognitive Sciences* 2014.

      Delete
  14. I found very interesting to learn that collaborative group remembers less than a nominal group. Do you think that this is due to a special organisation that able people to remember only a kind of thing and to count on others for the rest, maybe as some people are using computer to replace some part of their memory?

    ReplyDelete
    Replies
    1. Thanks Claudia. The standard explanation for the robust collaborative inhibition effect is that working in a group inhibits individuals' retrieval strategies - my access to the organization of information in memory is disrupted by having to attend to other people. Actually, though, many of these experiments operate with groups of strangers taking turns to recall material that isn't significant to them: we have tweaked this basic design in a range of ways that increase the 'groupiness' or interactivity of the process. So no, in general I think we only count on others to remember for us when we know them well.

      Delete
  15. I came across some interesting research on decision-making and visual attention recently that this work reminds me of. Orquin & Loose (2013) have an excellent review of the role that attention plays in decision-making, which requires mental resources such as working memory to hold all the relelvent information about the available options in a given decision. They describe the just-in-time attentional strategy whereby, individuals display an increased number of fixations for shorter durations, effectively relying on an external memory space when the task demands on working memory are high. This to me is a great example of a distributed cognitive process where environmental queues become part of the cognitive process, which is reliant on their relative stability.

    ReplyDelete