One purpose to rule them all....

Tweets are nice.  Getting a vibe for where people's thoughts are at is interesting, especially if you aren't working within a vibrant research team.

This tweet (part of an apparently upcoming debate between Pasi Sahlberg, Graham Brown-Martin and George Siemens) is particularly ripe for  deconstruction.

The main thing I find interesting is the possible hidden assumption that education's purpose is 

Image from http://sd.keepcalm-o-matic.co.uk

1. singular,

2. frameable in noun-like language,

3. stabilizable.

Now obviously I don't think a tweet reveals any of these assumptions.  However, these assumptions do tend to come across in amateur educational discourse, and even at academic level discourse.

1. Singular & 3. Stabilizable

The history of education certainly suggests some sort of cyclical (or perhaps non-deterministic) pattern.  This suggests it is highly unlikely that education will ever tend to a single stable "purpose".  I personally take the view that education's purpose is non-deterministic, rather, I assume that it can be described a a cycle between (two) characterizable states.  I'd also add that there is always a non-zero chance of everything going-to-pot (complex dynamics).

2. Frameable

The other issue I really wonder about is the assumption that education's purpose is frameable in normal language.  By this I don't mean that the process or end-point orientations aren't frameable, just that you can't easily separate out process & state.  This leads to a bit of a conundrum: 

• is it more effective to take a process approach, giving up a bit on end-point characterization ease and introducing functionalist bias, or
• take a fully functionalist process approach giving up the ease of talking about "things",
• take a fully non-functionalist process approach, like multi-level selection theory.

Mixing process & state is certainly unstable and enables intended and un-intended abuse. A fully non-functionalist approach is accurate, but perhaps unsatisfying for those that like descriptors (thing-talk).

Perhaps, when talking about education's purpose, the best we can do with common-language is to adopt the "tension" term.  Thus education's purpose is worded as some sort of tension between large-group orientation and smaller-group orientation.  As can readily be seen, leaving things this vague, while accurate, is unsatisfying.  Thus the fully non-functionalist process approach, while accurate, leaves one striving for more statist (thing-based) descriptions.

Here's a summary some ways to frame educational tension according to the three options to the stated conundrum.

Functionalistic Process	Process + State	Non-functionalistic Process
social equity ends	large-group orientation characterized by behaviour which elevates the value of social equity	expressed dominance of large  group traits
vs.	vs.	vs.
contextualized academic/vocational needs/preferences	small-group orientation characterized by behaviour which elevates the value of contextualized needs & preferences	expressed dominance of small group traits

What is obvious is that any process+state combination of explaining education's purpose probably requires more cumbersome language than the simple noun descriptions people seem drawn to.  Similarly people just aren't used to pure process descriptors.  Nor is a process approach economical in this case: there are just lots of (functionalist) sub-processes going on.

CONCLUSION

So, before debating what "schooling is for", it's probably best to talk about where along the process-state spectrum educational purpose can be adequately characterized for the depth of discourse sought.

The Cultural Evolution Piece

This summary of a paper by Henrich, Boyd, and Richardson is exactly what I needed.

Image from http://photos.demandstudios.com

While I have yet jumped very far into the cultural evolution literature, its a piece that is needed to support how (institutionalized) education functions as an adaptive group.

MODEL SUMMARY

My basic model posits tension in education between group-level orientation and individual-level (or more precisely, smaller group) orientation.  It also posits a slight preference for toleration.  Toleration preference orients the system to group-level inclusivity (i.e. toward a universalist moral position).  This fosters low frequency but sudden phase changes.  Reform resistance emerges due to the weak-to-moderate moral nature of the group.  Reform resistance is therefore governed by the various social factors associated with adaptive groups (i.e. norm variation detection & punishment, costly commitment displays, moral big brother, etc.).

I'm still playing with the notion of whether the end points of large group-orientation and small group-orientation are or are not strange attractors.

CULTURAL EVOLUTION CONNECTION

That's why this quote from Henrich, Boyd and Richardson is so interesting:

[C]ultural transmission does not involve the accurate replication of discrete, gene-like entities. Nonetheless, we also believe that models which assume discrete replicators that evolve under the influence of natural- selection-like forces can be useful. In fact, we think such models are useful because of the action of strong cognitive attractors during the social learning. 

The reason is simple: cognitive attractors will rapidly concentrate the cultural variation in a population. Instead of a continuum of cultural variants, most people will hold a representation near an attractor. If there is only one attractor, it will dominate. However, if, as seems likely in most cases, attactors are many, other selective forces will then act to increase the frequency of people holding a representation near one attractor over others. Under such conditions, even weak selective forces (“weak” relative to the strength of the attractors) can determine the final distribution of representations in the population.

Furthermore, the summary of their position by carcinisation.com is at least as interesting.

A major feature of cognitive attractors is that particulate cognitive information is less costly to hold and transmit than blended information - for example, it's easier to model the moon as either purely a rock in space or purely a conscious entity than some combination of the two.

Education's social equality camp is a cultural attractive well.  Education's academic/contextualized-needs camp is another cultural attractive well.  Intractability in premise or conclusion judgment prevent a stable solution.

Now whether the end-points are or are not strange attractors is a hard question.  Dynamics could be due to the attractors or due to the environment which the attractors lay (i.e. the judgment process).

??? Proposition-Conclusion = Individual-Group ???

Image from http://blog.colorjive.com/

Multi-level selection theory posits tension between group-level interests and individual interests.  In some cases group orientation is more adaptive than an individualistic orientation.  In lieu of worrying about fitness, List & Pettit (2011) take a philosophical (logic-based) approach based on List's judgment aggregation theory.  This is used to filter some of the conditions involved in judgement aggregation, bypassing the impossibility theorem (there is no single way to aggregate a single net-judgement when propositions are interconnected non-trivially with 3+ people)/.

List & Pettit's preferred way out of the impossibility trap,

involves prioritizing some propositions over others and letting the group attitudes on the first set of propositions determine its attitudes on the second. (pp. 56)

CONNECTIONS TO MULTI-LEVEL SELECTION?

Now what strikes me as interesting in this approach is its dyadic nature.  This is eerily similar to the dyadic nature of multi-level selection (which is usually operationalized with a two-level view).  While I may be stretching a bit, a little tidbit from List & Pettit is provocative: suppose the group assigns priority either to premises or conclusions... Further suppose the group delegates responsibility to premises to subject-specialists and to conclusions to the laity.  To me, at least, this roughly parallels multi-level selection's individualist vs. group tension.

In my nascent "Education as an adaptive group theory" I've suggested tension between group orientation and individual orientation models the dynamics of education's reform resistance and a slight preference for tolerance creates a 50 year total reform cycle.  However, observations hint that "individual orientation" as used by evolutionary biology isn't a perfect fit.  In education, the "individual orientation" seems to center around what is appropriate for contextually localized, identifiable small groups.

For example, I'd suggest that in education, an orientation for a system optimized for the small percentage of academically oriented critical thinkers is an "individual orientation".  In multi-level selection theory I currently suspect I have to posit 2 group levels within education: the supra "institution of education group" and the sub "special interests in education groups".  While likely feasible, this isn't exactly parsimonious.

INTERESTING FOILS

Image from http://www.extremetech.com/

What is interesting though, is thinking about whether special interest education groups tend to operate via premise or conclusion based judgements, and whether Education operates via premise or conclusions based judgments

Education has a historic tension that, since the mid-1800's has cycled between social equality and academic content progression.  I've tended to view the social equality paradigm as based on an universalist based avoidance gradient: i.e. "Do you really believe this person/group doesn't deserve a proper education?"  On the other hand, I've tended to view the academic paradigm as based on contextualized analyses flavoured by return on investment justifications: i.e.  "Let's do what is best for identifiable groups, making sure we scaffold the system for those smart cookies who'll make a real difference for society."


While my gut feeling suggest the social equality paradigm seems conclusion based and the academic paradigm is principle based, however, I suspect there isn't enough information to go on.  Plus I'm not sure how reasonable it is to try, at this level, to differentiate premises and conclusions.  My gut feeling is pretty much based on the hunch that the academic paradigm seems more granular/contextual, thus at a lower organization level, and thus more likely to be based on components rather than a whole. However, this reduction seems logically unrelated to the proposition-conclusion question.

UPDATE:
Interestingly enough, List & Pettit do go on to suggest that group agents can allocate premise based judgments to subject experts while conclusion based judgement can remain with the laity.  While this seems to match up with a common-sense intuition of how groups function in practice, it does nothing to get around the impossibility problem.



References

List, C. & Pettit, P. (2011).  Group Agency: The Possibility, Design, and Status of Corporate Agents. Oxford University Press

Defining Group Agents

Image from http://www.zazzle.com/

While other projects have come up this fall, I thought I'd return to my ongoing challenge supporting the theory that western institutionalized education functions as an adaptive group.

My muse has been strong work in the science of religion by folks such as Atran, Norenzayan and Wilson.  My background in this project has been informed by generalized complexity theory (Prigione, Kauffman, etc.) and social network analysis (Cross & Parker, etc.).  The initial, formal, trajectory was via biology / evolution / multi-level selection theory (Wilson, Okasha, etc.).  Since summer, I've ben trying to rationalize/compare this initial trajectory with social science work, particularly Sawyer's excellent summary work.  I'm finally getting around to the economics perspective via List & Pettit's Group Agency work.

List & Pettit's book excels due to the purposeful transparency of their logical arguments.  Theory frame-work with a philosopher co-author!  They're coming at the group agent question from a joint intention path.  As can be seen below (List, Pettit, 2011), they take some remarkably different approaches from the biology crowd:

But while cultural evolution may plausibly shape existing group agents, just as competition shapes commercial corporations, we are not aware of any examples of new group agents coming into existence among human beings this way.  It is hard to see how individuals, each with his or her own beliefs and desires, could be organized without any joint intention, or continuing intervention, so as to sustain and enact group-level beliefs and desires distinct from their individual ones.

Their conditions for jointly intentioned groups (i.e. their "group-level") are:

• Shared goal 
• Individual contribution
• Interdependence
• Common awareness.

As an off-the-cuff exercise, let's see how institutionalized education stacks up...

Condition	Definition	Ed's Fit	Reason
Shared Goal	They each intend that they, the members of a more or less alien collection, together promote the given goal	4/5	Salient goals vacillate between social equality and academic actualization. (note: vocational education straddles my binary).
Individual contribution	They each intend to do their allowed part in a more or less salient plan for achieving that goal.	5/5	Teaching is based on extremely high intrinsic motivation. "Doing it for the kids" is a quasi-religious-like mantra
Interdependence	They each form these intentions at least partly because of believing that the others form such intentions too.	2/5 to 5/5	Lot's of uncertainty here.  The disdain of for-profit colleges and private k-12's is telling.  Most educators believe students come first. The degree to which socialization influences intentions vs. self-selection of intentions is uncertain.
Common awareness	Each believing that the first three conditions are met, each believing that others believe this, and so on.	5/5	Education is remarkably monolithic in purpose (increase abilities) despite huge differences in goals (equal abilities vs. actualized abilities) and paths (public-private, content knowledge-critical thinking, etc.)

Overall, I'd say education looks like it should fare pretty well in List and Pettit's group agent framework.  While I'm still working my way through the book, I'd say education's crux is liable to be the condition of idea rationalization.

The third front on which a group must organize itself...[is] that whatever beliefs and desires it comes to hold, say on the basis of its members' beliefs and desires, form a coherent whole. (pp. 37)

Now, I'm not too sure why List and Pettit took this road.  People seem quite good at non-rationalization of belief.  In fact the rise of Haidt's social-intuitionist model (and various others), seem  to suggest too much focus on rationalization is a red-herring.  And maybe it's my recent biological bias, but it seems like hard rationality is much harder to defend that an adaptive level of rationality.  But then, I guess it's easy to excuse a philosopher's rational bias when they make theory so clean....


References

List, C. & Pettit, P. (2011).  Group Agency: The Possibility, Design, and Status of Corporate Agents. Oxford University Press

Tangents - Formative Assessment effect size: Content feedback or Socializer

The second tangent that's crossed my mind during recent research on another topic is about the causality producing formative assessment's high effect size.  Two recent posts over at Larry Cuban's always excellent blog are the muse (socialization & ed reform).

Is formative assessment's rather robust effect size mainly due to content specific feedback causes, or due to social/group dynamic causes?

As per Michele Kerr's post up on Cuban's blog, I don't think the power dynamics in a class which emerge in social interactions should be underestimated.  Formative assessment necessarily establishes the teacher's social presence.  Bandura's Social-Cogntive theory certainly suggests that changing group dynamics and environmental influences alters goals (individuals and group), which thus change performance.

Perhaps an easy way to verify effect sizes would be to compare the difference in formative assessment effect size when formative assessment is provided by

1. a computer known not to be controlled or influenced in any way with the teacher
2. written feedback from the teacher
3. media rich feedback from the teacher
4. one-way live feedback from the teacher
5. interactive two-way feedback between teacher and student

Results should be studied on a micro and macro time-scale.  Chances are pretty good that prolonged exposure to 1 or 4/5 will make a significant difference in social dynamics in play during teaching.  Obviously I'd have to done some much better thinking in order to figure out how to control for teacher-student/class sociality with 1.

If anyone knows of studies which have gone down this path, let me know...

Tangents - Rationalization of Instructional Pieces

While working on a research paper about a design-based professional development (PD) initiative a few tangential thoughts have crossed my mind.  Here's the fist.

Alignement/Rationalization of Instructional Pieces

The first was reasonably germane to PD: is effective teaching usually due to alignment/rationalization of various nested instructional pieces? In other words, does having an aligned/rationalized

1. educational philosophy, 
2. delivery method, 
3. instructional process (i.e. assessment, classroom management, mastery expectation, expected student effort, etc.) and,
4. instructional strategy

correlate with effective practice (as either measured by students or teachers)?

My suspicion is that it certainly does.  The small sample qualitative research I've been doing suggests that instructional process is extremely tacit and opaque.  It also suggests that not many teachers conceive of adjusting, let-a-lone actually adjust, delivery methods.  The Carnegie unit and "bums-in-seats" mentality are robust.  Educational philosophy is generally pretty easy to probe and easy for teachers to describe practically.  Instructional strategies are highly contextualized by experienced teachers.  Changes in instructional strategy really only happen when they align with educational philosophies.  After all, education is loosely coupled, and even the most firmly managed instructional strategy changes quickly become hybridized.

Economically poor, low performing schools

If institutionalized education functions as an adaptive group, one implication is that it should be fairly resistant to change.  Successful adaptive groups typically aren't swayed by potential freeloaders looking to do things that may only be good for themselves or a small sub-set of the population.  Now, this doesn't mean innovative behaviours aren't permitted, only that real, fundamental innovation relating to group function and moral mission is highly unlikely.

A corollary to this is that populations subject to lots of change, especially fundamental changes relating to moral mission, may be on the peripherally of the adaptive group or not part of it at all.

This leads to an interesting point: in the US, it seems like low performing schools in economically disadvantaged areas undergo significant reform efforts.  In practice this often means hopping from one reform flavour to another, often very quickly.  In many schools, such reform practices are mitigated and hybridized.  It seems to me that in these poor school forces exasperate rather than mitigate reform pressure.

Is it therefore reasonable to conclude that such schools and such students are on the periphery of education's adaptive group?  If so, will it take a major phase change back to a state more intolerant of reform experimentation to bring this population back into the adaptive group fold?

In other words, while reform efforts are intended to help out disadvantaged populations, on an evolutionary, do they do the opposite by risking population exclusion from the adaptive group?

One of the cruxes of this idea is whether the changes in poor schools are fundamental to education's moral mission or are merely superficial tactical changes.  I don't know.

Application Test

Over the summer my education reform resistance ground-work seems to have settled down.  At least to me, its seems seems to have come together as a larger framework that I was expecting.  In fact, it is probably best articulated as a full fledged theory: Education as an adaptive group. While there is lots more work to get it constructed as a theory, perhaps it's time to test what utility the dynamical model associated with this idea has….

The Organisation for Economic Co-operation and Development (OECD) recently posted a regression of innovation in education system innovation vs. 8th grade math teacher satisfaction.  The results show a moderately strong positive correlation.  A conclusion is that teachers are more satisfied when education systems innovate more.

SITUATING THE TEST CASE

This research fits with Bogler's (2001) finding that teacher prefer to work with a leader that exhibits transformational behaviour rather than transactional behaviour.  It also fits in with miscellaneous research on human preference for novelty, which certainly seems to be an evolutionary selected trait.  Of interest to me, the OECD finding also fits, tangentially, with Sigmund's & Nowak's (2001) simulation which shows minor preference for tolerance among individuals leads to increasing levels of tolerance leading to cyclical transitions to extreme intolerance.  Of course, to apply Sigmund's work you have to make a rather large leap in assuming increased tolerance also means increased novelty/innovation.  Such a leap is fine for armchair bloggers, but certainly isn't academically sound.  Nonetheless, it seems like the OECD finding is well supported by what I know (but haven't cited) of evolutionary cognitive science research, educational leadership research and teacher preference research.

RE-CAPPING BIG IDEAS

Education as an adaptive group theory (or the current nascent underpinnings of this potential theory) suggests the institution of education, and those who are considered members of this group, are dynamically torn due to competition of between-group (group) selection and its proximate causes and within-group (individual) selection and its proximate causes.  Weak dominance of one factor over another creates instability (Okasha, 2009) while strong dominance creates stasis.

An adaptive group, such as institutionalized education in rule of law states, is subject to moralization selection. This includes things associated with freeloader detection, norm variation punishment, and transcendental moralization (Atran, 2002; Wilson 2002; Norenzayan 2013; Haidt 2013).  Institutionalized education is a moderate moralized group.

A functionalist interpretation of the education-as-an-adaptive group model postulates a variety of different attractor descriptions for the group end state and for the individual end state.  Each pair of descriptors should be complimentary (i.e. tolerance & intolerance would pair as a dyad, but curiosity & power would not).  Education functions, amongst other things, as a societal coherer that provides significant fitness benefits for its constituents.

A non-functionalist interpretation of the education-as-an-adaptive group model postulates tension between group processes and individual processes.

In both the functionalist and non-functionalist interpretation, group processes/states dominate weakly.  Continued group expansion or maintenance of large group stature leads to increased heterogeneity. This can be envisioned a the rise of individual processes (non-functionalist interpretation) or attraction to the individual state (functionalist interpretation).  This eventually challenges adaptive group function.  A phase change occurs to a homogeneous state/process (adaptive group state).  The cycle repeats.

Education's competitive relationships are fairly stable in scope: its sphere of influence is not rapidly changing.  Group competitors may include such things as homeschooling, private tutor based education, non-institutionalized religious education, religion, some radical outcome-based charter schools, and perhaps political groups.  K-12 competitors differ from post-secondary competitors.

VALIDATIONS

Relevant to the OECD report correlating increased innovation with increased teacher satisfaction, education as an adaptive group theory accurately predicts findings.

1. Teacher preference for innovation - Homogeneity of the group state is unstable.  Individual process &/or sub-group splintering rises complexly.  Processes that facilitate individualization should be favoured.  Education's moderate moralizing characteristic minimizes the degree to which it favours extreme conservatism.  Education has just the right degree of freedom to facilitate innovation but prevent group implosion.
2. Education is near or slightly above the average innovation levels of other professional sectors - Institutionalized education has functioned as an adaptive group for 150 years (since the emergence of popular public education).  Education's moderate moralizing characteristic minimizes the degree to which it favours extreme conservatism or extreme progressivism.  Education has just the right degree of freedom to facilitate innovation but prevent group implosion.  

PREDICTIONS

Here are a few predictors I came up with which can't be verified from the data.

1. 50 year period for innovation preferences - Education as an adaptive group model predicts that teacher preferences for innovation should wax and wane with a major period of 50 years.  Every 50 years one should see a spike in in-tolerance toward divergent practice.  A correlated decrease in innovation preference is likely.
2. Minor (complex) cycling - Within each country, one should expect to see minor cycling between high and low levels of innovation preference.  Tension between group state and individual state produce complex dynamics.  Greater dominance of group processes or individual processes should produce more stable dynamics.
3. High moralization facilitates divergence or extreme conservancy - As the severity of education's moralizing character increases one may tend to expect either a shift to extreme conservancy (i.e. religious like behaviour), or the ability to function with increased levels of divergence (i.e. the moral mission is strong enough to unite disparate individuals).  
4. No moralization leads to unsustainability - As eduction's moralizing character vanishes one may expect to see education not function as an adaptive group (i.e. degrees and accreditation may have little value). One may also see unsustainable institutionalized operation at the group level (i.e. lots of total reforms & co-optation of purpose).

CONCLUSION

The main finding of the OECD piece, preference for innovation, seem to weakly validate education as an adaptive group theory.  However, there is a fair bit of 'handwaving' involved.  Functionalistic argumentation is very weak. While I've been trying to stay out of the functionalistic trap, it seems likely I'll get caught in it in order to explain why the group operates with a background bias for toleration/innovation.

Sigmund's work shows a slight preference for toleration leads to the cyclical dynamics observed in education.  Wilson's (and others) recent work on altruism will likely provide some ammunition supporting minor preference for inclusivity/tolerance.  However, more work is obviously needed to show that education doesn't function in a highly competitive environment, and thus doesn't require tight norm detection and punishment.  I suspect some of the work on the emergence of universalizing religion would be quite informative here.  For example, Beneke's book on the origins of American pluralism is likely to be quite good.

Additional work on education's role as a large group is needed.  For example, does a large group role require universalizing tendencies such as slight preference for innovation, divergence and tolerance?  Again, this seems to be a likely draw into functionalism.

REFERENCES

• Atran, S.  (2002).  In Gods We Trust: The Evolutionary Landscape of Religion.  Oxford Press.
• Bogler, R.  (2001).  The influence of leadership style on teacher job satisfaction.  Educational Administration Quarterly, 37, 662-683.
• Haidt, J.  (2013).  The Righteous Mind:  Why Good People are Divided by Politics and Religion.  Vintage.
• Leach, C., Zebel, s., Vliek, L, Pennekamp, S., Doosje, B., Zomeren, M. Ouwerkerk, J., Spears, R. (2008).  Group-Level Self-Definition and Self-Investment: A Hierarchical (Mulitcomponent) Model of In-Group Identification.  Journal of Personality and Social Psychology,  95 (1), 144-165.
• Norenzayan, A.  (2013).  Big Gods.  How Religion Transformed Cooperation and Conflict.  Princeton University Press.
• OECD (2014).  Measuring Innovation in Education: A New Perspective, Educational Research and Innovation, OECD Publishing.  Retrieved August 20, 2014 from here.
• Okasha, S. (2009).  Evolution and the Levels of Selection.  Oxford.
• Sigmund, K., Nowak, M. (2001).  Tides of Tolerance.  Nature, 414, 403.
• Wilson, D. (2002).  Darwin's Cathedral: Evolution, Religion and the Nature of Society. University of Chicago. 

Quick Summary

Here's a quick summary of my approach to an explanatory model of the education reform problem.

1. Posit that the institution of education functions as an adaptive level group as characterized by Wilson's multi-level selection theory.
2. Model the tension (boundary points / attractors) which occurs with between-group selection (group) and within-group selection (individual).  Weak between-group dominance produces the complexity and dynamics of education reform resistance
3. Interpretation of this modelling can be done within a functionalist paradigm or within a non-functionalist paradigm.
1. A functionalist interpretation investigates what drives motion to the two complexly competing attractors.  It also investigates possible attractor descriptions.  It assumes  a reductionist paradigm/philosophy.
2. A non-functionalist interpretation investigates nesting among groupings and competition between groups.  It assumes a process philosophy/paradigm.

One feature that comes out of this model is stochastic complexity characterized by cycling between the group state and the individual state.  Sigmunds (2001) Tides of Tolerance model is a good fit.  This model shows that minor individual preference for tolerance causes a slow but steady rise of tolerance in a modelled population followed by a rapid phase change to intolerance.  Behaviour is cyclical.