A great overview of the story of the Cognitive Sciences up to the current state of affairs. It serves wonders as a recap of The University of Edinburgh’s “Introduction to Cognitive Science” course and more. Clark is a great philosophical writer, very witty and clear, and verging on the poetic, which is how I strongly believe the human mind should always be referred to.
Amongst the many things it does, one of its foremost achievements is justifying the Cognitive Sciences as a unified field:
“The scientific study of mind thus demands interdisciplinary effort and multidisciplinary cooperation on a whole new scale, probing adaptive response at multiple organizational levels including those incorporating bodily, cultural, and environmental scaffolding.”(Clark, 2014, p. 249)
And now with regards to what it has done for me in terms of elucidating certain concepts and inspiring new questions.
Moving backwards in a sense, from my most recent to oldest thoughts or realizations, “Mindware” has made it clear that while PP and extended/embodied cognition can be argued to work together as ‘friends, not foes,’ (Clark, 2014, p. 243), they are also fundamentally, at least if considered radically, opposing in their nature: PP involves a representational computational architecture, while (radical) embodied/extended cognition does away with any need to represent the world, using the ‘world as its own best model’ (Rodney Brooks’ famous slogan). Note that this discussion of the relationship between Em-Ex-cognition and PP is still under debate. We might only go so far as saying that it is a ‘love hate’ relationship.
Yet this relationships emerges only if the goal of the organism is “supporting self-organization” (p. 243)….which is something that I’ve often pondered. On the one hand, it is not clear what exactly constitutes self-organization. On the other hand, it seems to me as if this tends towards describing human agents as cognizing based on the very ‘Fristonian’ purpose of maintaining the necessary states for existing. But…how much of our cognition, especially higher-level cognition, involves a preoccupation with our instinctive drive to survive?
But to err on the positive instead of the suspicious for a hot second, I’ve always wondered about how certain inventions of today, like the Macbook for example, seem so ordinary and intuitive to us now that it’s absurd to think that computers used to take up an entire room (!!). PP, however, demands that this be the case. If we assume that the brain is mainly a predictive engine of gladiator-stature and layered like the Colosseum, then our cognition can only possibly evolve through a time-elapsing process of “neuro-socio-cultural unfolding” (Clark, 2014, p. 245) where we sculpt and structure in ways that afford new possibilities to ever-incoming waves of generations of humans. In this way “we train new minds to predict novel, artificially generated patterns” (Clark, 2014, p. 245). No wonder, then, that the musical genre of rap evolved from the AABB rhyme schemes of Kurtis Blow, to the infinite, unpredictable multisyllabic, internal rhymes of Eminem and other lyricists, and Breakdance evolved from simple 6-step footwork to incredible feats of gravity-defying, suicidal acrobatics. PP demonstrates, to extend the Roman metaphor, that Rome was not only NOT built in one day, but neither by just one Emperor, or better yet, neither by the prisoners of only one war.
Related to this is that PP favors the neuroconstructivist vision where the world constructs the patternings of our brain-connectivity in a way that affects our interactions with the world, which in turn changes the stimuli that the world presents to our brains, and the cycle repeats itself. This closely resembles the theory of ontological design expounded by Anne-Marie Willis and introduced to me by Jason Silva which is roughly (and I say roughly because this is how it has been paraphrased to me, not because I am summarizing the paper which I have admittedly yet to read) that our actions affect our current state and pave the way for certain (as opposed to other) future states to ensue.
This brain-body-world processing story, that looks more like a tangled bush than a sequence, coupled with the feedforward and feedback loops involved could be the very key in describing what makes humans humans and in explaining the source of worldly experience that is so particular to us! (WHY it has to feel like anything at all, though, is still up for speculation…that’s the hard, or comically, XXX problem).
I’d like to note, in addition, the suggested tie between PP and metacognition on p. 244, where a meta-model might be required to pick between what I like to call the two Players in PP: our inner models vs the sensory data. In other words, metacognition might be heavily implicated in our day-to-day, moment-to-moment decision between utilizing a computationally-heavy inner model to tackle the world, or be lazy instead and offload the work unto the world, where we increase the weight on the sensory information in such a manner that our strategy is thus “knowledge-sparse,” (Clark, 2014, p. 244).
This was surprising to me. Up until now, I believed that what was actually ‘lazy’ of us was our tendency to depend on our models of the world as opposed to focusing on the data in front of us. When I think about it, it seems a lot easier to stick by a pre-defined model than to maintain a continuous awareness of the incoming flux of information. Isn’t that the reason why models are being postulated in the first place? Such that the only data that requires ‘brain-power’ is the data that diverges from the norm? Isn’t the supposed easiness of leaning on models the reason why somebody is more likely to have an accident in their neighborhood than elsewhere (Burdett, Starkey & Charlton, 2017) because they are less alert to potential changes in an environment that experience has made them concretely encode in a specific way such that any such change is too rare to care about, resulting in ‘inattention blindness’?
Maybe I’m onto something…or maybe there is something I’ve poorely understood.
Nevertheless, PP seems to offer a wonderful landscape of opportunity, although I’m late at declaring this because a lot of that opportunity has already been acted upon. The PP framework and everything that has led to its coining has, despite being a relatively new approach, been years in the making. A lot of these years were spent in the underground, in the sewers of science, with different researchers and thinkers arriving at similar conclusions in the dark, not knowing that the network they were all scuttling about was that of PP (note what I said earlier about the role of building over time and over multiple generations).
This is a crucial time to begin wondering how PP can help us become better people and live better lives. This is where philosophy should direct its efforts moving forward, I argue.
In addition, maybe in a more hyperbolic vein, could PP help us enhance our skills faster than previously possible? For example, assuming that we will eventually not only understand the exact intricacies of the hierarchical generative model and the predictions/corrections that occur at each level, but also master it…predict it (although Clark shows how predicting may be tantamount to understanding on p. 236), could we become better, say, dancers or athletes or argumentators or lovers? By simply, in other words, better understanding and eventually controlling the interplay between sensory data and representations that guide everyone’s cognition? Isn’t that already what magicians and illusionists/mentalists do? Maybe a conscious, exact mapping of the activity at each level of the prediction-stratum isn’t even necessary, if not pretty much impossible. The world is too complex for that level of knowing. And yet, we might still use the knowledge of that structure to our benefit. Before we think about becoming superhuman by artificial means, it might be worth harnessing the ways we can becomes better humans by exploiting our own innate faculties. This might help guide us later on (and not so far off) when AI and robotics becomes entangled with our biology. Cue the ethics of AI in 5, 4 ,3…
PP is already on its way to help explain processes that, although we are very well acquainted with, are very hard to explain intuitively in real time without the framework. If I were to ask you to explain how reasoning and planning is possible – how it is possible to envision things in the future and plan accordingly – you might end up saying that…well…it just is the way it is. But without PP’s down to earth, nitty gritty, structured explanation, planning actually resembles shamanism more than anything. Imagination is no different, probably even more so. Clark relates these phenomena of our cognition to PP in p. 240 and p. 236 respectively. That thinking, predicting and acting can turn out to emerge from the ‘same broad computational apparatus’ (Clark 2014, p. 240) seems monumental to me.
Though is PP really so down to earth ? Clark mentions in p. 238–239 that PP has yet to be backed up by empirical evidence, especially its architectural subdivision into a representational level and error-encoding level. This is very important to note and keep in mind. We must now acquire this empricial evidence, or explain why it can’t be found, at least not so easily. Personally, I’d imagine that if PP is argued to be so pervasive, then shouldn’t evidence be abounding as opposed to scarce? Maybe…in which case, the lack of empirical support for the framework would be worrying…or I might just not be reasoning correctly. As Sherlock Holmes often points out, it is the most seemingly evident and simple cases that turn out to involve the densest web of complications and the most amount of work to uncover the cues.
Yet, going back to the potential role of PP in our experience of the world, Clark claims that “to the extent that our guesses capture the evolving sensory data, we perceive the world,” (p. 236). This makes me wonder: where do we pinpoint the source of experience? In the error signal? Later on in the corrected prediction? Or previously in the uncorrected hypothesis? In the worldly data? Somehow in all of these channels and their interactions and more? I feel like Clark would argue for the later and also scold me for asking irrelevant questions, but that makes me wonder about where do we draw the threshold? What minimum amount of activity and dynamic interactivity is required before actual conscious experience materializes? Or is ‘materializes’ the wrong word if we are to concede that “perception is controlled hallucination” as per the dictum of Max Clowe? Well, the doors of perception (#Huxley) are just as open as my questions regarding the matter.
How much is PP and error-consideration implicated in serendipity and ‘aha’ moments?…etc, etc.
The questions keep coming forth, and they are fascinating ones. Yet, what I’ve laid out to you here is only with regards to my thoughts around Predictive Processing which, might I add, is only spoken about in Ch. 11 of Mindware! This is simply to demonstrate how thought-provoking Clark’s exposition of the developments that have defined the history of Cognitive Science really is. I reiterate that it is incredibly informative and yet easy to read, with countless readings suggested at the end of each chapter for further self-indulgence.
Cognitive Science is, in my opinion, the most marvelous of fields. And it is thanks to authors and thinkers like Clark alike that we may have a paper transcript for why Cognitive Science is not only “the new kid in school” but the “badass new kid” who is not ‘too cool for school’ but quite the contrary.
Clark, A. (2014). Mindware: An introduction to the philosophy of cognitive
science (2nd edition). Oxford University Press.
Burdett, B. R. D., Starkey, N. J. & Charlton, S. G. (2017). The close to home
effect in road crashes. Safety Science, 98, 1–8.