Tuesday, 5 August 2008

Environment and self organisation

As you’ll have gathered, I’ve been pre-occupied recently with self organisation, self regulation and learning. Thus two phrases leapt out at me when my Planet project colleague, Yishay Mor, blogged about Europlop 2008.
“Funny thing, how social scientists will stand before a hall full of neatly packed passive listeners and preach about collaborative constructivist learning, while the computer scientists just do it.”

“EuroPLoP 2008 also leaves ample space and time for the real work of conferencing, the stuff that can’t be captured in predefined structure, because it emerges out of the discussions at the time and place.”

For Yishay, then, working in a research context, the really valuable process is self organisation rather than regulation (whether by self or others). I note, though, that for the outcomes of self organisation to be useful research ideas the environment has to be structured in an appropriate way – which implies a fair degree of control or regulation of the environment and an understanding of the types of environment that encourage effective emergent behaviours - or, in memetic terms, a knowledge of the types of environment that will select useful behaviours. So, for the conference to be a success, the environment is controlled although the entities within it are not. Perhaps, though, even here the environment is less controlled than at a conventional conference? Is deliberate absence of control, in itself, a form of control?

I’m not entirely convinced by the memetic, selection, interpretation. It is not clear to me why a relatively uncontrolled environment should select useful research behaviours. It might select all sorts of other things (eg. a lot of alcohol addiction, if Yishay’s emphasis on beer is to be believed). There is evidently a much more complex relationship between environment, interactions between conference delegates, and attributes of the delegates than a simple selection model would suggest. Perhaps, an uncontrolled environment, a community of research practitioners (ie. the interactions) and delegates with “academic” attributes are all necessary conditions. If these conditions are independent, then one would say that selection of the right sort of delegates, and community of practice, had happened before the conference started: only those who meet the necessary conditions are in jobs that will pay for their attendance – they have been ruthlessly selected through educational and career structures. However, it also seems entirely likely that the conditions are not totally independent - that the environment, community and delegates help bootstrap each other up to the useful research behaviours. This, after all, is presumably why we encourage research students to attend conferences.

In a previous post, I suggested that for effective learners, the attractors, or stable states, are educationally desirable behaviours. It seems to me that there might be two ways we could ensure this. The first is by the environment ruthlessly selecting those who have these behaviours – this approach might be said to characterise HE prior to the mid 1960s and the beginning of a widening participation agenda. The second is by helping learners to develop strategies for finding these states, and for avoiding other, equally stable but educationally undesirable, states (for example by turning them into repellers). If my analysis of Yishay’s conference experience is near the mark, then a self organisation model of learning needs to understand, not only the necessary environments and communities for learning, but also the dependencies of environment, community and student attributes that will help students become learners.

Monday, 4 August 2008

Thinking about transitions between self organisation and self regulation

In a previous post, I noted that Carver suggests that self regulated learning and self organised learning are two different processes, and that different parts of the brain may be responsible – although this is, he says, a speculative proposal. If it is true, though, we cannot have a linear transition from one to the other – the process cannot reside in some intermediate part of the brain. We have the question then, of whether the change from one to the other is a discontinuous flip – perhaps once the self organised learner has found an attractor and reached a stable state, this triggers the self regulated process and switches off the self organising process. Gestalt psychology, and experience of "Ah ha" moments might suggest that this is the case. Alternatively, the change may look continuous – as the learner homes in on an attractor the self organising process may gradually tail off, with the self regulated process gradually increasing, and the overall learning process represented by the convolution of the two.

Under what conditions could self regulated learning take over from self organised learning. It seems to me that the essential condition for self regulated learning is to be able to conceive of the goal and a process for reaching that goal (?charting). Feedback then comes into play in monitoring the process and progress towards the goal. To be able to conceive the goal and process one must have some experience of it already; this is why I do not see how cutting edge research can be self regulated. But the experience may be vicarious, and it is this, I think, that makes self regulated learning in an educational context possible.

We also have an issue of goal definition. Does the process of reaching a goal help to define the goal? Or is a goal that we do not know how to reach, necessarily one we consider ill-defined? Or has this more to do with whether the measure of achievement is an internal subjective or external objective measure?

Consider the goal, “I want to get a 1st class degree classification”. This, looks like a pretty well defined and objectively measured goal. If I am the child of upper middle class, university-educated and employed parents, and am at a university surrounded by similar peers, with a close academic relationship with tutors and a well scaffolded course, then I have a pretty good conception both of what that goal means in terms of performance, and of what I have to do to achieve it. My parents, and many of my friends, and tutors have all done it previously and provide a model, or pass on tacit knowledge through practice, in Harry Collins’ terms.

If, though, I come from a disadvantaged background, with no parental experience of university education, and go to a university with a high number of other widening participation students, then the goal seems less well defined, despite the objective measure, since I am likely to have far greater difficulty of conceiving what it feels like in terms of the my performance, and even less of how I go about achieving it, however good the tutor and scaffolding on the course may be. Monitoring of both the process and the performance becomes difficult. I have to trial or sample different behaviours to find one that seems to be working, and my learning initially looks more like self organisation than self regulation. Even when I have found one, I don’t know whether it is a tried and tested behaviour that leads to the goal, or a potential dead end, so I am less likely to stabilise on that behaviour.

In the latter case, I am much more akin to someone who has what seems an ill-defined goal, eg. “I want to be happy”. I can conceive of being happy, but have no objective measures of happiness. I know a number of happy people, but they all seem to have achieved it in different ways, and there seems no fixed route to happiness. The best I can do is work on a day to day basis, making decisions that I hope will promote my happiness, and hope that the emergent state will, indeed, be happiness.

We can, I think, conceive of these issues in complex system terms by saying that self-regulation is only possible within the basin of attraction of an attractor. Social links with, and vicarious experience of people who have already achieved the goal and are already in the attractor, are a strong (?necessary) condition of being located within the basin. Thus, in social terms, the difference between one’s ability to self organise and to self regulate lies in one’s distance from others who have already achieved the goal: the widening participation student is likely to be using the same self organising processes as the Nobel Prize winning researcher – neither knows exactly what they are doing – both are self organising.

The requirement for proximity to someone who has already achieved the goal, ties in well with activity theory and the zone of proximal development. It also provides a role for the teacher, even in learner centred learning design and participatory, socially constructed learning – the teacher is someone who has already achieved the goal. If they expose their process of achieving the goal they can help attract learners into that behaviour ?an apprenticeship model of learning.

Key Reference:
Carver, C. S. (2004) “Self-Regulation of Action and Affect” in Baumeister, R and Vohs, K (eds) Handbook of Self-Regulation, New York: Guildford Press, pp13-39

Sunday, 3 August 2008

Memetics & Self Organisation

When Anoush alerted us to Susan Blackmore’s talk on memetics it sparked a flurry of debate. Like the rest of the CA I have a number of issues with Blackmore’s argument. However, what I was struck by was the relation between Universal Darwinism, and the idea of self organisation discussed in my last post.

In my snowflake example of self organisation, we have, essentially, two factors at work: the interactions between the water molecules, and the interactions between the water molecules and the environment (largely defined by temperature and pressure)

It seems to me that Universal Darwinism provides three factors as a basis for self organisation:
Mimicry/replication/heredity – ie, the interactions between entities
Selection – ie. the interaction between an entity and its environment
Variation – ie. an autonomous characteristic of the entities

If the entities are genes, we get the whole panoply of living species that we know, as emergent states – from amoeba to armadillos.

If the entities are practices, we get memes.

And, according to Blackmore, if the entities are information, we get temes.

If one takes this self organisation view, it becomes clear how, as Blackmore claims, living species, memes and temes can emerge bottom up and without any top down purpose or direction.

It also seems that the Universal Darwinism example captures the three fundamental types of factor we have to take into account when thinking about any self organising system: the characteristics of the entities; interactions between the entities; and interactions of the entities with the environment

Saturday, 2 August 2008

Self Organisation versus Self Direction and Self Regulation, and Learner Autonomy

I (Isobel) have been reading some of the psychology literature on self regulation recently, and also following up on the (rather underdeveloped) psychology ideas about self organisation - leading to this over-long blog post. Sorry folks.

Self Regulation has its foundations in cybernetic control theory. There are a large number of variants on the idea, but core features are that:
• Processes are purposive
• Self-corrective adjustments are made to adhere to the purpose
• The corrective adjustments originate within the person
The basic metaphor is of feedback control with four elements:
• Input function
• Reference standard (defined by the purpose)
• Comparator
• Output function
Behaviour is a process of moving towards goals, or away from anti-goals, via feedback control.

Self Direction. So far I have come across no definitions in the psychology literature (although of course there are definitions in other literature). For linguistic reasons, though, I would take it as differing from self regulation in two fundamental ways:
• The purposes or goals are set by the self (in self regulation they may, but need not necessarily, be set by others)
• No specific mechanism for processing towards the purpose is proposed
I take it as similar to self regulation in
• The primacy of goal setting (both in importance and chronologically)

The cybernetic control metaphor underlying self regulation and (perhaps) self direction is well defined and has clear experimental implications when considering one goal, or factor, at a time. On this view a successful autonomous learner would have the ability to plan an efficient route to their goals, and have effective feedback mechanisms to keep them on the route. The focus of research would be on planning competencies and feedback mechanisms.

However, it seems clear that the complexity of the cybernetic control metaphor increases rapidly (?exponentially) as the number of interacting goals increases, and rapidly becomes unmanageable, especially if the algorithm connecting the goals is undefined.

Self Organisation has its origins in dynamical systems theory. One aspect of dynamical systems theory is the mathematical theory of complex systems of interacting factors. Key ideas are:
• Emergent states: the form and properties of the whole cannot be predicted from the component parts but arises from the interactions of those parts and without top-down purpose eg. snowflakes – each has a different form, and they cannot be predicted from looking at an individual water molecule – they arise from the interactions between water molecules without anyone/thing/top molecule organising all the rest. They are self-organised.
• Phase transitions: states are stable against small perturbations, but a slightly larger perturbation in one factor may trigger a change to a completely different (emergent) state of the system. For example a snowflake at -5C is stable against changes in temperature of up to 5, but as soon as the temperature rises even slightly more than this, the flake suddenly melts into water. For some systems an infinite number of phase transitions or states is possible.
• Attractors and repellers: Some configurations of the system are stable against small perturbations – these are attractors. If it is near such a configuration, the system will self organise into such a stable state. Similarly, some configurations will never be adopted and the system will not spend any time here – these are repellers. If it is near such a repeller, the system will self organise so as to move away from this state. In between are configurations that are not stable, but that the system may move through. For example, a pencil lying on its side is in an attractor. A sharp pencil will not stand on its point (repeller), but it will move through a number of leaning positions as it falls on to its side.

Such ideas have been applied to subjects from maths, physics, chemistry, to the stock exchange, social systems and psychology. From a psychology/learning viewpoint, self organisation seems a candidate metaphor for systems where there are:
• Multiple interacting factors
• Changing situations
• Emergent states
• Lack of purpose
• Lack of predictability

Carver (2004) views attractors and repellers as analogous to the goals and anti-goals of self regulation. To be successful as a self organised learner, it seems to me crucial that the attractors are educationally desirable outcomes, while the repellers are undesirable outcomes. In practice most learners are surrounded by undesirable as well as desirable stable states (truancy, addictions, low level computer games, etc). Hence, perhaps one thing that characterises successful learners is their ability to convert undesirable attractors into repellers. Another characteristic would be their ability rapidly to find and configure themselves around desirable attractors: the learner will move into a state where they can “do this”. We could, for example, model this by thinking about an individual moving through different configurations defined by cognitive, affective and social parameters. It seems to me that, assuming that at first they try different configuration solutions by sampling, there might be several ways in which they might locate and move towards the attractor more rapidly than another learner. They might sample more rapidly – eg. switch attention rapidly from one possible outcome to another. They might sample at a greater distance from their current configuration – eg. by greater cognitive leaps. They might have a greater number of sampling points – eg. by a bigger social network (although note that they might also have to switch attention more rapidly to take advantage of this). They might have a stronger pull toward the attractor once found – eg. by greater interest, recognition or reward. There is likely to be an element of randomness in all these processes.

Putting it another way, one might posit, for example, that successful learners can synthesise a large number of information/skill sources, keep a broad overview of multiple possible outcomes, are sensitive to the proximity of a stable state, and able to strengthen the connections that will move them into that state if it seems a desirable one.

Carver suggests that individuals perform both self regulation and self organisation, and that different parts of the brain may be responsible. He suggests that self organisation is the appropriate process in new and undefined situations, but that as a pattern of behaviour emerges and is recreated, it becomes familiar and defined and self regulation takes over. He cites evidence that different parts of the brain are involved in learning a new skill from repeating an established one.

On a larger scale, human systems are capable of both self organised, and regulated behaviours. Liu et al (2005) draw an analogy to patterns of traffic on a motorway composed of all the cars which are more or less autonomous within the broad constraints set by the environment. These represent a self organised system. They differ, for instance, from the very rehearsed and regulated patterns of an american football team

It seems to me that in the CA we are studying two different scales, and it is worth distinguishing between the two:
• Organisational scale. Here the organisation (HE institution or commercial company) is the system, and the people within it are the entities of which the system is comprised.
• Individual scale. Here the individual is the system, and its cognitive, affective, social predispositions are the entities of which the system is comprised

In the self organisation/complex systems approach, the description “autonomy” is applied to the entities of which the system is comprised. At individual scale, then, it does not make sense to talk of an “autonomous learner” because “learner” refers to the macroscopic behaviour of the system (the individual), while “autonomy” refers to the microscopic entities (cognitive predispositions etc) of which the individual is composed. It would make sense though, to describe the behaviour of a company, as a system, comprised of autonomous learners as entities. For example the emergent, self organised, state of a company comprised of autonomous learners might be a learning organisation (if the company was lucky:-))

It would, though, make sense to talk of an individual as a self organised learner. In this case the entities of which the individual is composed, have organised themselves into a behaviour which we would describe as “learner”.

At individual scale, then, one might look as self organised learners, their abilities and behaviours. Alternatively, at organisational scale one might look at the types of institutional emergent state that arises from different behaviours and interactions of individuals.

Summary of differences between self regulation and self organisation

Self regulated Self organised
Purpose No purpose (maybe
Top down control Bottom up organisation
Autonomy characterises the unit Autonomy characterises the constituent entities
Clearly defined goals No goals or poorly defined goals
Best when dealing with a single goalCan accommodate multiple interacting factors
Constant environments Changing environments (maybe)
Predictable outcomes Unpredictable outcomes (maybe)

Key references:
Carver, C. S. (2004) “Self-Regulation of Action and Affect” in Baumeister, R and Vohs, K (eds) Handbook of Self-Regulation, New York: Guildford Press, pp13-39

Carver, C.S. and Scheier, M.F. (2002) “Control processes and self-organization as complementary principles underlying behaviour” Personality and Social Psychology Review, 6, 304-315

Liu, J., Jin, X. & Tsui, K.C., 2005. Autonomy-Oriented Computing (AOC): Formulating Computational Systems with Autonomous Components. IEEE Transactions on systems, man, and cybernetics - part A: systems and humans, 35(6), 879-903.

Hooker, C. & Skewes, J., 2007. Abstract of “Complex Systems Dynamics: Implications for Sustainability, conception and policy.” To appear in Gabbay, D., Thagard, P., and Woods, J. (eds) Handbook of the Philosophy of Science. Elsevier. Available at: http://www.johnwoods.ca/HPS/Hooker_Skewes_abstract_v2B%20F.pdf [Accessed August 1, 2008].