AdamWTeach

Load to Prediction - Variation Theory In response to Christian Moore-Anderson's question on how Predictive Processing / Active Inference theory applies to aspects of Variation Theory

Please find my response ... i'm a bit prone to verbosity so had to do a blog!! #EruditePedagogy

open.substack.com/pub/predicta...

..... thinking.... I am taking the time to read your book ( Difference Maker) so I can better address your excellent question, thinking of making a concrete example, e.g. using your speaking Russian and then discovering foreign languages.... I struggle with brevity, so need to publish as a blog.

However, it links to PP in terms of the brain evolving as a prediction machine. The "forgetting" mechanism, the design of the biochemical cascades and messaging, the effect of dopamine, pyramidal neuron structure are all the evolved structure that is able to infer salience from predict. errors *6

I agree that PP/ActInf does not feel to give clear explanatory support to why retrieval practice and spacing are effective. I believe that comes from understanding the neurochemical processes around how engrams- networks of neurons -synapses, receptors, dendritic spines etc. all actually work. *5

In addition the fact that desirable difficulty and managing limits to learning load can fit together with the same neuro science backed mechanism fits an occam's razor principle for me *4

re. Desirable difficulties, my point is that "difficulty" is subject to interpretation, and lots of work has gone into trying to pin it down in terms of element interactivity etc. but I find the PP model of large enough surprise to trigger model updating compelling and backed with neuroscience. *3

... If PP/Act inf is used as the base cog architect, rather than lim. working mem. the level of prior knoweldge as the major effect is a direct outcome. in fact complexity of task design was also another factor, but this is displaced by student familiarity (predicticability) of the task design *2

The point here was the Expert reversal effect was a sticking plaster to explain why "cognitive Load" didn't increase with element interactivity. In fact findings were that prior knowledge was the main factor in the self reported feeling of "load" - *1

#EruditePedagogy .. let's start... @cmooreanderson.bsky.social you felt there were some areas of variation theory that Predictive processing /active inference theory don't address/explain/support... What are your thoughts here? I like to understand and develop that thought.

@cmooreanderson.bsky.social I have started a Note on substack as a place to develop this discussion on variation theory - I would like to understand and discuss which parts of variation theory you feel that PP/ActInf don't explain well. You have more extensive knowledge on Var theory than I do!

Interesting stuff ... "Organisms that predicted threats or rewards had survival advantages. Brains evolved not to process information, but to minimise surprise and prediction error. Over time, this machinery generalised across all sensory and motor systems."

I li think the combo of pp/act inf and the neurobiochemisty will help but not be the whole picture of the mechanism by which the brain updates it's generative model.. how does it infer statistical patterns and generalities? This may well link to and be informed by why variation theory works

When I say guts of it I mean the system/ fundamental mechanisms, that the effects, instructional theories and pedalogical strategies sit upon. But I think we also need to go the layer lower into the biochemical processes of synapses and neurons to understand more on how learning happens.

I've skimmed the difference maker, and agree with the fact that there have been many of the ideas and philosophies that have come before that have likely been some cornerstones, steps towards, supporting acts to PP/actinF. It's my feeling that PP/ActInf is getting to the guts of it all.

In fact Christian you comments on variation theory being an instructional theory vs. prediction processing being a theory of cognition helped shape this blog.

Christian, many thanks for your comments, I had the feeling that this is where your comments thus far were coming from.. I embrace them and have already started a blog in responding... I.e. furthering the socratic devabate

Change Load to Prediction?
open.substack.com/pub/predicta...
@tombennett71.bsky.social @olicav.bsky.social @karlfristonnews.bsky.social @profcoe.bsky.social @pepsmccrea.bsky.social @walkthrus.bsky.social @hugheshaili.bsky.social @evidencebased.education @krisboulton.bsky.social

New episode!! 🎙️📣

A conversation w/ Dr. Gordon Burghardt about play across the tree of life.

It's not just kids, apes, and dogs that play. Play abounds—often in unexpected places. It takes different forms, has different functions, and still presents puzzles.

Listen: disi.org/the-primeval...

Challenge accepted..

I'm prepping a new substack on it now. Just trying to balance nerdiness to accessibility!

On the point of engrams and neuronal firing I have done a deep research dive in how our brains do store /update it's information (we'll update it's generative model). Is fascinating, but also informs spacing/retrieval and has some unique implications for teaching I think

But I do think that it's incumbent on us, and especially on those thought leaders to take note of and understand if there are new and improved theories of how our brains work, and how best to help them learn to best work in the future world our students are going into.

In teaching I am sure the likes of @adamboxereducation.bsky.social tom Sherrington @teacherhead.bsky.social Craig Barton, Ollie Lovell, Sarah Cottinghatt etc who are in lots of schools would say the biggest bang for the buck is in basic behaviour / good practice.

I achieved some the best yield/quality improvements by doing the simple stuff, but over time the theoretical stuff gave some step changes in capability. I think you need both.

Yup, takes me back to my days working in engineering on quality improvement teams in the hard disk drive industry. You could take the theoretical path understanding deep physics to improve things, or you could work with the production line on eliminating simple basic errors

Also, I started on this with it understanding autistic and other neurodivergent learners. Also I understand sweller, willingham etc all all admit that their models are not intended to be theories of how the brain works, just useful models to guide good teacher practice.

I'm noticing that it's now making its way into physiology/mental health increasingly. Some big recent papers on understanding depression in this lense and it finessing strategies to address it.

The working memory limit 4 items, 7 items, "it depends", it's contextual, it changes iif it's a story, it depends if you can aggregate lumps of knowledge into memorable objects => prediction error bandwidth

Yes but, when a new fundamental theory simplifies and addresses many complexities in current theory then it becomes useful. Like the whole determination of "load"- element interactivity, difficulty,' germaine load, prior knowledge etc. = predictability.

I think pred processing active inference theory is a better model of how the brain works. But would it's understanding/use drive even better teaching practice? I think it could be like general relativity to Newtonian physics or any other new scientific model transition, it takes time to see impact