DJ Strouse

My Goodreads Rating: 4 of 5 stars

Interested in cybernetics, theoretical biology, and philosophy but still find Dan Brown novels to require mental gymnastics? Put on your philosophical training wheels and give “Tree of Knowledge” a spin!
A mixture of dated scientific ideas, profound frameworks for thinking about living organisms, and unnecessarily complicated jargon, ToK is essentially the children’s menu version of Maturana and Varela’s Autopoiesis and Cognition papers on living organisms, communication, and consciousness.

I highly recommend reading ToK before Autopoiesis and Cognition and possibly even foregoing Autopoiesis and Cognition altogether. ToK is not only more clearly written but is laden with examples, something lacking in the uncompromisingly sterile Autopoiesis and Cognition.

The rest of this review is a summary of the deep and profound wisdom I gleaned from the Chileans, so you may want to skip it if you haven’t read the book yet.

ToK’s more gentle approach (along with post-reading conversations with a Chilean economist and Italian physicist) helped clear up a question I had after Autopoiesis and Cognition: if a unity is so deeply coupled with its environment, how does one uniquely define its morphological boundaries? It may seem obvious to look at me, carve a 2D surface over my skin, and call me a closed system, but give me a week without a consistent supply of low-entropy energy and I’ll quickly succumb to the second law of thermodynamics. The key trick is this: unique boundaries there are not. “Everything said is said by an observer.” An observer selects the features by which a unity will be defined through their shared domain of interactions. Different observers (and even the same observer at different times with different goals) will have different domains of interactions and will define a unity in a different way. For example, a given university may be a set of assets and liabilities, a collection of students, a football team, a physical space, or some combination of these things, depending on who you ask.

Some more notes:

• Referring to a unity implies an act of distinction.
• Replication, copy, and reproduction can be distinguished by the amount of historicity in each process. Replication (repeated generation) is ahistorical. Copy (creation from a mold) is historical if iterated. Reproduction (the fracture of a unity to create two unities of the same class), however, is necessarily historical.
• Heredity and variation are strongly complementary features. Heredity is the preservation of structure in a historical series of unities. Variation are the differences of structure in that series. Different components of a unity may exhibit different degrees of heredity and variation.
• Unities may couple via inclusion (think organelles) or recurrent coupling with the maintenance of individual identities (individual humans).
• The environment does not instruct an organism; it only triggers internal dynamics. To phrase it differently, the space of possible reactions to an environment is defined in the internal structure of an organism; the environment does not inject behavioral commands into an organism in any way. To phrase it differently yet again, environmental stimuli modulate, they do not control. Environmental input is imply one more “voice” in the “conversation” of internal dynamics.
• Organisms must exhibit variance of the time scale of their environment (and in a complementary “direction”) in order to adapt (remain coupled).
• Adaptation in response to a single change in the environment affects the organism in a global way. A small change in structure may occur to accommodate one new feature of the environment, but through an internal domino effect, alter the way an organism interacts with other features.
• The simplest neural systems allow detection of correlations between inputs on a sensory surface.
• A nervous system expands our possible behaviors by inserting a network with a huge range of possible patterns between our sensory and motor surfaces.

View all my reviews >>

Tags:   cognition · cons · cybernetics · neuroscience · psychology · theoretical biologyNo Comments.

“What are you gonna do with your life?”

College freshmen (and many graduates… and many adults) all grapple with this question. If you’ve been naive enough to answer “science!” then you’re faced with the follow-up: “Well, what are you gonna study?”

Most young knowledge warriors charge straight at whichever field their best teacher from high school or college was from, certain that their initial passion will never wane. Yet talk to these firecrackers seven years later and notice how many grad students have lost that spark. What once was a tantalizing realm of puzzles and mysteries becomes a never-ending cycle of mundane tasks.

Science can certainly be a cruel succubus, but considering one important feature of a field before hopping on the bandwagon can help avoid this.

Ripeness & Saturation: The Potential of a Science

The key is to compare the current set of ideas with the potential for developing new ones given available data, tools for acquiring data, and culture. Too little new available data, too few tools for acquiring data, or cultural norms that enforce old paradigms can all hinder the potential for interesting projects. Quantum gravity might be an instance of the first two shortages whereas 16th century astronomy (influenced by the church doctrine of a geocentric universe) is a good example of the last. Let’s call these saturated fields.

On the flip side, an abundance of data, the invention of tools for acquiring new data, or changing cultural norms can all provide opportunities for great projects. Genomics is experiencing the first two surpluses whereas complex systems research in the early years of the internet (and the metaphors about networks that accompanied it) probably benefited greatly from cultural shifts. Let’s call these ripe fields.

Your mission, young sailor of the scientific seas, is to find one of these gems.

Now, how can you tell? You can’t be sure, but here are some good signs:

• Researchers still identify themselves with another more traditional field and just list this one as an “interest.”
• It attracts researchers from many different backgrounds and much of the work done is considered “interdisciplinary.”
• A new institute pops up every week or so.
• A large percentage of new papers all reference one or a few papers written in the last five years.
• The introductory textbooks were all written in the last few years and there is no consensus on which is best.

Some signs of a saturated field include:

• Systematized jargon (new ideas and discoveries get ID numbers rather than names)
• A large percentage of new papers are written to make small corrections on old ones (the field has become recursive).
• There are introductory textbooks in their twelfth edition and most professors use the same text to teach as they did to learn.

I’d wager quantum computing and neuroscience are ripe for the picking.

Which fields would you add?  And which criteria for identifying ripe and saturated fields would you use?

Tags: 2 Comments

My Goodreads rating: 4 of 5 stars

Wow, I’ve never felt so mentally humbled in the shadow of a biologist. In the realm of arrogant physicists and mathematicians, biologists are seen as the housewives of science – keeping things clean and tidy while the real men do the work. I’ve met enough intelligent biologists to know that this is only the case most of the time, Maturana is a giant. I feel no shame in admitting that this was one of the most difficult books I’ve slogged through and that I’d often spend 10-15 minutes on a single page. That said, it was worth the slog.

My reactions to this book are a mixture of the following three three-letter phrases: “wow!”, “duh…”, and “wtf?!” The “wow!”s were accompanied by large-scale synaptic migrations as my paradigms regarding life and cognition were scrambled. The “duh…”s were my response to Maturana’s incessant repetition of ideas only a Baptists alligator wrestler from the Deep South would argue with – evolution is a blind and local process, biological systems are recursive, blah, blah, blah. This might, however, be as unfair as accusing Shakespeare of adhering to every stereotype in Western literature, as I’m pretty sure Maturana was an early pioneer in the still fledgling field of theoretical biology and that many works I’ve read since are derived from his ideas. The “wtf?!”s were in response to Maturana’s needlessly complicated lexicon of undefined terms. It seems like he and Varela went off and lived in a forest for 20 years, shielded from civilization, and developed their own strange and impenetrable vocabulary that only they understand.

The “wow!”s occurred almost exclusively during the first essay of this book: “The Biology of Cognition”. I was much less impressed by “Autopoiesis”, probably because the central idea of this book, recursion, has since spawned a closet industry of books ranging from masterpieces of human thought to crackpot theories on how Gödel’s theorem proves that God invented the internet.

As usual for books that woo me, I’ll reserve my fifth star for another few weeks/months to see if my infatuation with the ideas in this book is nothing more than a teenage fling or something truly special and lasting.

Finally, the following are the main ideas I drew from the two essays. These notes are mainly to aid my aging memory, but you’re free to treat it as a poorly executed synopsis. My criticisms of the text follow afterwards.

• Cyclical (Autopoietic) Systems
  • A living organism is a cyclical system whose pieces provide for their own synthesis and maintenance (call this process “autopoiesis”).
  • The disruption of this cycle destroys the organism.
  • This cycle relies on the environment; it continually makes predictions about the environment by requiring and expecting certain resources. If these predictions fail, the organism may die.
  • One goal of an organism is to expand its environmental requirements (and thus predictions) into broad classes rather than very specific conditions. In this way, the organism becomes more robust to environmental change.
  • These cycles (autopoietic systems) may be nested, smaller cycles being the components of larger ones. There may even be level-mixing in which interactions play roles on multiple levels.
  • There is some wiggle room in which an autopoietic system can be perturbed and yet still carry out its autopoietic self-genesis. That wiggle room constitutes the cognitive domain. It is the space of biological deformations that do not destroy an organism.
  • As autopoiesis defines an organism, the relations between the components that constitute that organism are far more important than the components themselves.
  • Organisms are fundamentally ontogenic. Development is not a process that culminates in an organism. The organism is the entire spatio-temporal pattern that includes development.
• Domain Distinction
  • An organism’s niche is not a subset of the environment an observer describes. The niche is defined in terms of the organism’s domain of interactions with its environment. The observer necessarily describes the environment in terms of his own domain of interactions. This is a major barrier to explanation and understanding.
  • An organism may interact with its environment in ways unobservable to others.
  • An organism may (perhaps dysfunctionally) interact with its environment in ways unobservable to it, but observable to others.
  • Communication is the orienting of one organism to a particular internal state by another organism. Note that the cognitive domains of the two organisms are different, so it makes no sense to speak of “information transferred” in the absolute.
  • Absolute denotation of communication exists only in the mind of an observer who notices a relation in his simultaneous interactions with both organisms.
  • Two organisms may only communicate if their cognitive domains have significant overlap. Otherwise, they are incapable of orienting one another to corresponding appropriate internal states.
• Neural Systems
  • Only that which leaves a signature on the nervous system may enter the cognitive domain. That which does not affect the brain is invisible to the organism.
  • Interactions that leave the same neural signature are indistinguishable to an organism, be they between the organism and its environment or between internal cognitive states. It is possible, however, that an external observer may be differentially affected by similar interactions and be quite capable of distinguishing them.
  • Neural systems can give a representation to “pure relations”, expanding the cognitive domain to include abstract ideas. With this, pure relations may begin to independently interact with one another.
  • Interesting view of a neuron: spatial system of possibly overlapping affector and collector areas
  • Neural systems function in the present. The past only plays a role to the extent that it leaves a signature in the brain that carries on to the present. In general, for the past and predicted future to play a role in cognition, they must be abstracted and represented.
  • The brain is local in interaction but not representation. Computation proceeds physically via matter affecting matter (interaction is local). Ideas, stimuli, and other neural states are distributed across the brain (representation is not local).
  • Internal states represent spatiotemporal interactions with an organism’s sensory service and subsequent internal activity.
  • There are at least three time scales to consider:
    • Immediate – stimuli transiently affect neural activity
    • Lifetime – (repeated) stimuli more permanently affect the organization of a neural system (learning)
    • Evolutionary – evolutionary pressures affect the “base genetic model” that prescribes an organism’s development
  • Neural systems change continuously and non-predictively. For a system to evolve between two states, the intermediate states must be accessible and viable.
  • Interesting domain in which to study neurons: the I/O domain
    • Fix I, vary parameters, and watch O change
    • Fix I/O, examine reduced parameter space that preserves that particular I/O relation
  • Questions
    • What are the fundamental units of the nervous system? What are the fundamental units of any information-processing system? That is, what should we treat as primitives in order to explain what neural systems do?

That said, the 40-year old essays do contain some outdated material, namely the (oft repeated) doctrine that neurons are deterministic. Neurons are not deterministic. Their input-output mappings are pretty friggin’ stochastic, owing at the very least to the fact that channel dynamics dip into the quantum world of chemical reactions.

I also suspect that the reason Maturana and Varela resort to such a tangled web of undefined jargon is that many of their ideas are less developed than the Olsen twins (warning: my bag of pop culture references has not been replenished since the mid-90s). First, how exactly does autopoiesis define unique topological boundaries for an organism? If the autopoietic cycle that defines an organism is so deeply interwoven with the environment, how does one separate organism and environment? Every organism relies on its environment for resources. How to draw structural boundaries is obviously much clearer to Maturana and Varela than it is to my feeble brain. Second, Maturana and Varela stress that our descriptions of the functioning of organisms are fundamentally flawed due to the domain distinction problems mentioned in the notes above. Why is their description of autopoiesis immune from these mistakes? Why are they so certain that autopoiesis is the definitive characteristic of life when they argue throughout the text that the true character of organisms is forever unknowable in our restricted cognitive domains?

View all my reviews >>

Tags: 1 Comment