Saturday, April 25, 2015

Radio 1: Consciousness as a Streaming Information Service

The benchmark of modern theories of consciousness, by Bernard Baars, 1986: "A cognitive theory of consciousness."

One of the more contentious and mysterious areas in science is consciousness. Its existence was long denied by behaviorist psychology, yet long-standing schools of (idealistic) philosophy have held that it is the only thing that exists at all- that all else is an illusion, or at least subsidiary to the great reality which is consciousness. The universe itself is conscious!

Well, that is absurd, but it indicates the trickiness of a subject that attempts to explain our ability to think about subjects. One quickly gets into philosophical, if not logical, thickets. But it is pretty clear that inanimate matter doesn't think, while we and our fellow animals have, after billions of years of painful evolution, graded abilities to think about things, including ourselves, which arise from that organ behind our eyes ... the brain.

The study of consciousness is now a hot academic field, but that happened only in the last couple of decades, after the long hiatus of behaviorism. Bernard Baars's book from 1986 is a watershed in the field, presenting a functional theory of consciousness which has little to say about its physiological basis, but has profound things to say about its phenomenology, internal logic, and purpose. Its model (consciousness as global workspace) remains the basis for current work in the field.

In some respects, consciousness is extremely small, not to mention slow. We can barely attend to one thing at a time and remember maybe five to seven things in short term memory. Anything we do consciously must be done at a snail's pace, and only once learned becomes faster as it also becomes automatized, and sinks into the unconscious. The unconscious is, in contrast, endlessly vast, taking care of physiological functions all over the body, analyzing speech as we hear it, translating our thoughts into speech as we speak it, and on and on. It is a parallel, not serial, fleet of processors, some of which are very fast. Any action we take is made up wholly of unconscious mechanisms. All we are conscious of is maybe the goal of reaching for a cup, (or typing a letter on a keyboard), and all the calculation and activations in between are taken care of, like magic. A practiced typist won't even think about reaching for individual letters anymore, but will fluently type by the word, or more.

Yet consciousness does one thing that none of these other, learned, unconscious processes do, which is broadcast information extremely widely over a vast population of other (unconscious) processors. When reading some text- consciously, and only consciously- it is judged by a variety of processes that operate outside of immediate consciousness. Is the syntax correct? Is the style fashionable? Is the content interesting? Is the spelling correct? Is the meaning connected to the last sentence? The fact that these issues and many others can only be analyzed when one is "paying attention" has great meaning for the nature and role of consciousness.

I used to think that consciousness was the caboose on the train of thought. The classic experiments of Libet showed that for any action, even so-called voluntary action, electrodes can pick up activating signals well before one is conscious of a choice being made. And it is obvious with the slowness and high-level nature of consciousness that anything that enters it has gone through a great deal of prior processing at other levels of the system. But that doesn't mean that consciousness is only a spectator. No other process provides the integrating, broad reach across virtually every process in the brain, at a high level. Bringing something into consciousness means testing it for coherence at many levels, from spelling to consonance with our model of the world and hierarchy of goals.

Thus consciousness is an active function, specially tasked with broadcasting novel data far and wide over the multifarious pool of unconscious processes, which can each in turn comment on what they see. Is an object moving in the distance? Such an event exites special visual processors and calls to attention what may have not been there before. Conscious attention then allows us to consult our full cognitive battery of memory, world model, goals, etc. to decide what that object might be, in relation to our needs.

This leads to Baars to consider the stream of consciousness phenomenon. Why is it so tenuous and fluid? The fact is that we are virtually unable to attend to unchanging stimuli. Even loud noises, if repeated endlessly, fade out of awareness. The hedonistic treadmill is notorious for habituating us to any pleasure or good forture we may experience, which is soon taken for granted and cast aside in a search for the *next great thing. Consciousness concerns itself with information, in the formal computational sense. Whenever some activity starts to pall or become routine, it fades from consciousness. When learning new skills, this is a good thing, as sufficient repetition causes the whole process to go automatic and unconscious, like touch typing. Good or bad, the phenomenon seems to be universal as well as clearly adaptive. We are constantly on the hunt for novelty and information. Whatever is been-there-done-that is relegated to the dustbin of memory, or obliviousness.

It should also be obvious at this point why conciousness is so narrow and singular. If some data is supposed to be broadcast and commented on by many, indeed all available, other processes for integrated evaluation of an ambiguous or novel situation, there can only be one such item at a time. Consciousness is a radio station we listen to one story at a time. It is, and must be, a serial processor. Baars marvels at the then-trendy studies in biofeedback, which demonstrated that with enough training, virtually every part of the body and mind can be manipulated consciously. Individual muscles can be trained, blood pressure changed, etc. This phenomenon is coming into clinical use for various prosthetic appliances, which can be controlled by all sorts of muscles or thought patterns that wouldn't at first glance be candidates for actuators of voluntary action.

What gets access to consciousness? Again, unconscious processes lead the way. A big driver is our internal hierarchy of goals, one of the major unconscious functions that interacts with conciousness. Do we want a sandwich? A walk? A newspaper? Money? Survival? Life is complicated that way, as needs come up all the time at all levels. When a super-high level goal comes under fire, we are rivetted. Earthshaking experiences change conciousness itself, by re-arranging the unconscious contexts that underly it- it is not the earth that shakes, at least most of the time. Hallucinogens like LSD can have this kind of deep, long-lasting effect. On the other hand, if information comes along that violates a high-level model of reality particularly egregiously, we may also block it out rather than take on the problem of breaking down our painstakingly developed models of reality, (contexts, in Baars's teminology). Thus we fight cognitive dissonance to save them by rationalization, confirmation bias, denial, etc. Bears makes a point of the coherence that conscious contents must have to be useful, which rests on the cooperation of the many unconscious processes / contexts that join to create conscious contents. If these sub-processes conflict rather than cooperate, the result may not be conscious novelty, but rather indecision at lower levels and lack of consciousness on that topic.

Dreams naturally arrive on this train of thought as well, though Baars leaves that topic as an exercise for the reader. Baars does note that consciousness is heavily visual, with much less vivid access to abstractions than to scenes in front of our eyes. He speculates that this may indicate the evolutionary history of consciousness began in straight sensation, and only later attracted goal evaluation, retrospection, planning, inner speech, and all the other aspects that so enrich consciousness for us. During sleep we are released from the immediate layers of the goal system, so one might hypothesize that the consciousness apparatus turns towards free experimentation, using imagery to explore the deeper levels of our unconscious contexts- the goal, memory and prediction systems, which are susceptible to many complexities and internal contradictions. Dramatic role-playing, which is such staple of waking entertainment as a way into the mysteries of the human condition, are here personally staged for our continuing development.

Indeed, the theater is a leading metaphor for consciousness itself:
One can compare the mind of a man to a theater of indefinite depth whose apron is very narrow but whose stage becomes larger away from the apron. On this lighted apron [i.e. front of the stage] there is room for one actor only. He enters, gestures for a moment, and leaves; another arrives, then another, and so on ... Among the scenery and on the far-off stage ... . unknown evolutions take place incessantly among this crowd of actors of every kind, to furnish the stars who pass before our eyes one by one, as in a magic lantern.
-Hyppolite Taine, 1871

Lastly, one must wonder at how this is instantiated physically. The fact, and it does seem to be a fact, that consciousness involves wide broadcast of its narrow content, with far-flung systems both understanding what they are receiving and sending back ongoing commentary, requires a lingua franca of the mind. One aspect of this is clearly the gamma oscillation, which co-occurs with attentive consciousness and fits the model of something that unites broad but temporary coalitions of brain areas. But what is the code? What is the wave carrying? Is mere coordinated activation from one specialized area to another enough to form useful communication? That is hard to believe, but as yet, our tools are too narrow, or too crude in time or space resolution to figure out what this code might be.

Baars speculates that all goals and abstractions seem to have fleeting imagery in consciousness, again harkening back to the perceptual bias of consciousness. Our use of metaphor in language is naturally tied up with this phenomenon. Thus it may be imagery that is in some functional sense the lingua franca of the system, thought that still does not say how so many parts of the brain could recognize this language, especially the many humble parts that really do not seem to deal in imagery at all, like syntax checking, posture, "aha"- type solution verification, etc. Perhaps imagery forms a high-level language of aspirations, fears, planning, etc., without being needed for low level processes. This would tie in with the Jungian conception of the unconscious, with its core of archetypal images, our experience of them in dreams, and our need to develop and express them in art.


  • Cognitive nature and nurture in Scrabble.
  • But is the presidential campaign "news" enough to enter consciousness?
  • People do have their own facts.
  • On anger, and its cheap dismissal.
  • Myths of the mythically-minded.
  • A big key to poverty- violence, lawlessness, corruption.
  • Like in Qatar.
  • Best case scenario from the WSJ: wages are awful, peaked in 1972.
  • The labor market won't do it alone ... we need better policy towards redistribution.
  • Do we need "low cost, low wage" economies anywhere in the US? Where public assistance makes up the difference? No is the short answer.
  • IBM luvs Louisiana.
  • Reason prevails, barely, in Comcast merger.
  • Props to the Hubble.
  • Neonicotinoids even worse than thought.
  • Islam: "But it also brought together peoples who’d never had a common worldview, or shared humanity, before." And.. "And there can be no freedom if we are stuck believing in people, like Hirsi Ali and her ilk." ... who had to move to the US due to Islamist threats on her life. The author is scattershot in his apologetics, but well-intentioned.
  • Please don't use bar graphs for complex data.
  • A little mesmerizing cymbalon.
The WSJ agrees- wages have been stagnant for a very long time in the US, clearly not related to productivity growth.

Saturday, April 18, 2015

Evolution Sweeps Away Diversity

Natural selection carries off a large portion of neutral genetic diversity in large populations.

One would expect that large populations accumulate much more genetic diversity than small ones, over time. But if you watch those nature shows about herds of wildebeest roving over the Serengeti, it is very hard to see that variation. They behave as one, and look highly similar. Indeed, contrary to naive theory, larger populations tend not to have proportionately more genetic diversity than small ones. Why? The classical equilibrium law of population genetics assumes that larger populations naturally would have more variation, proportional to the number of members and the lengths of their various separate lineages. To balance this out, it also takes longer for any single new mutation to spread through such a population, so the ultimate rate of fixation of new mutations is no faster in large populations that it is in small ones.
"Under the assumptions of the neutral model of molecular evolution, the amount of variation present in a population should be directly proportional to the size of the population. However, this prediction does not tally with real-life observations: levels of genetic diversity are found to be substantially more uniform, even among species with widely differing population sizes, than expected."

But empirically, this expected high level of variation has not been true, even for neutral (unselected) alleles. This difference between theory and reality has been termed a paradox, and a recent paper (review) recounts the arguments above, showing that it is natural selection which constantly clears off accumulated variation, including completely neutral alleles that have no selective effect at all. This paper is not the first to address this whole paradox theoretically, but is the first to give an definitive quantitative solution.
"We show that genomic signature of natural selection is pervasive across most species, and that the amount of linked neutral variation removed by selection correlates with proxies for population size. We propose that pervasive natural selection constrains neutral diversity and provides an explanation for why neutral diversity does not scale as expected with population size."

Comparison of two species, one with large population size (fruit fly, A) and one with very small population size (Przwalski's horse, B). The gray dots are estimates drawn from 500k basepair windows across each genome of the local recombination rate (X axis), which can vary a great deal along chromosomes, and the local level of mutation and variation (Y axis). In a completely neutral theory, these measures should not correlate with each other (red line). The model developed in the paper is shown in the blue lines, where in a large population with lots of selection going on, regions with relatively low recombination show dramatically less variation, consistent with the rare selected mutation in those areas (whether positively or negatively selected) carrying a large number of neutral alleles with them, either to fixation (positive selection) or to their demise.

The issue is one of linkage. Imagine a long chromosome, with lots of genes and mutations. If one of those mutations is bad, then all the other mutations near it will be carried along with the bad one into oblivion, even if they did no harm themselves. The degree of linkage is a matter of the local recombination rate. Some areas of our genome recombine much faster than other areas, and thus allow more fine (selective) separation between nearby mutations, as they end up in different gametes and individuals due to the recombination that happens during meiosis.

So these researchers took a census of multiple genomes from many different species, (63 billion sequencing reads in all), measuring local recombination rates and mutation rates. They found that the bigger the species' population, the more clearly the prediction of correlation between the two measures came out in the data. Thus fruit flies, with a vast natural population, have roughly two-thirds the genetic diversity one would naively expect. The rest seems to have ended up shot down, innocent victims standing a little too close to more deleterious mutations.

In smaller populations, selection is just as fierce, but the level of neutral genetic diversity isn't expected to be as large in the first place, so loss by random drift plays a stronger role than loss as a byproduct of selection.

Humans are an good example. Now we are a huge population, but in genetic terms, we are practically clones compared to most other species. This is mostly because we were a very small population not long ago, and have only reached seven billion in an evolutionary eyeblink. So we have the genetics of a small population. But even in small populations, selection will have this diversity reducing effect, at a lower level. The intense selective evolution we went through over the millions of years prior not only kept populations small, but spread attractive and advantageous features through the population, at the expense of some of the other variation that was lying about.

In a way, this is an explanation for why species remain coherent entities through time. Their genetic diversity doesn't just grow endlessly into genetic chaos, but stays centered, in some abstract sense. Recombination and mating keep the genetic elements of the population continually mixing in a cloud of closely related forms, but it is selection that trims the outliers, both neutral and deleterious, keeping the cloud coherent, even as it also moves the entire cloud in new directions over the evolutionary landscape.



  • The progress of inequality. (with graphs). Did supply-side mean 1%-side?
  • Some sharp words for those new atheists.
  • But people will believe anything. In for a penny, in for a pound with Scientology.
  • A hopeful sign towards a more equitable world.
  • Theology remains utterly absurd.
  • Keynes on inequality, interest, the lower bound, and demand.
  • Pay what you wish: the IRS is now toothless.
  • "Redistribution", or justice?

Saturday, April 11, 2015

RNA, RNA Everywhere

The enhancers that drive transcription are themselves transcribed, in a regulatory process.

The last decade or two have not only brought a genomic revolution in molecular biology, but also remarkable discoveries in RNA, finding micro RNAs, conserved long non-coding RNAs, piRNAs, siRNAs, snoRNAs, and now eRNAs, for enhancer RNA. Even though most of the genome is junk and remains junk, 80% of it is transcribed, so the cell turns out to be a flurry of all sorts of incredibly diverse RNAs beyond the classic molecular biology trinity, which is: mRNA to carry the gene sequence from the DNA, tRNAs that serve as the plug-in adapters between triplets on those mRNA messages and the amino acids they will become in the protein, and the rRNA that forms the body and catalytic core of the ribosome, operating the converyor belt that brings together the first two RNAs to synthesize proteins.

In retrospect, we perhaps should not have been so surprised, since RNA has been there from the most ancient period of life, and the messiness of biology tends to elaborate complexity, using any wrinkle or handle as a regulatory process. But for a couple of decades we were blinded by the preponderant relative mass (and, to be fair, importance) of the RNA trinity in the cell, and only recently have we had the technical means to find the great diversity lurking beneath.

A recent review catalogues the findings and hypotheses about the newest member of this tribe, eRNA, in detail. In eukaryotes, especially as they become more complicated, genes are driven by quite elaborate collections of "enhancers", which are DNA segments typically far upstream, by thousands to hundreds of thousands of base pairs, that bear a cluster of DNA binding sites where regulatory proteins bind, which either turn that gene off or on. One gene may have many separate enhancers, each typically devoted to one phase of development and/or one location in the body where it drives the activity of its target gene.

Schematic of gene control, showing an enhancer (LCR) that has several colored regulatory proteins bound to it. At the same time that it loops through space to contact its target (ßmaj gene), it is also transcribed to short RNAs (red) by RNA polymerase (P). The small discs all over the place are histones (H), which are modified with various colored methyl and acetyl groups in another regulatory process.

Enhancers can do this because they form loops from their distant sites, to contact the start point of their target gene, at what is called the promoter (pictured above as a bold elbow+arrow, when active). This arrangement means that it hardly makes much difference how far away the enhancer is- the proteins it binds can ignore the many kilobases, sometimes hundreds of kilobases, of linear distance in the DNA between themselves and the target gene's start site. But it also means that there needs to be some way to "insulate" one gene and its gaggle of far-off enhancers from those of other genes, which one wouldn't want crossing over into each other's territory and turning each other on. That is a story for another time.

The new and quite paradoxical finding is that enhancers are themselves transcribed, and that these resulting eRNAs are not just accidental junk, but play a significant role in the operation of the enhancer and the regulation of its target gene. As pictured above, (in red), eRNAs come streaming off the enhancer long before the target gene gets turned on. And if those eRNAs are degraded by an experimenter's intervention, typically (and ironically) by programming siRNAs against them, then the target gene turns on much less than otherwise. So it is not just the act of enhancer transcription that is important, though that is thought to have some regulatory effects as well, but the products themselves, at least in some cases.

eRNAs are thought to interact with another level of regulation, which operates through the histones which typically package all eukaryotic DNA. Any protein that binds to a specific site needs to get through this packaging, which can happen in some cases by detecting the DNA on the outside of the histone, or by waiting for a stochastic loosening of the histone from the DNA. But after the pioneer proteins find their sites, they can attract other regulators that specifically modify lysines (K) on the histone with methyl and ethyl groups, neutralizing their charge and lowering their binding affinity to the negatively charged DNA. This process "opens" up the chromatin for other regulatory proteins to bind. The specific lysines that are modified on histones constitute a complex code that marks areas in chromatin for various stages of transcriptional and other activity. The eRNAs have yielded mixed behavior in this pathway, sometimes being required for histone modification at target genes, though not typically at the enhancer region.

Much is still unknown about these eRNAs- how general their occurrence is, how they work, what these little RNAs are doing in the enhancer-promoter complex, and what drives their own transcription. It is like wheels turning within wheels, within wheels- where does the gene activation process ultimately begin?


  • Bonus reference on eRNA.
  • The NCAA competition is wonderful, but its organization and inequality are not. This should be nationalized.
  • Like lots of other things.
  • Bibi has a screw loose.
  • Yes, religion doesn't make any (rational) sense. And, yes, theological institutions are a farce, educationally.
  • In case you were clueless about the NBC saga.
  • Pilots are another abused class of worker. No wonder one gets depressed.
  • Austerity correlates with recovery ... negatively.
  • Open carry? Not at the NRA convention.
  • People are instinctively socialist, and fair. Image from the talk, on inquality:

Saturday, April 4, 2015

Our Drug War: Ignoring Social Poverty and Exporting Paranoia

How our horrible drug war became everyone's nightmare: Johann Hari on Point of Inquiry.

Unusually, this is a podcast review, rather than a book review. But you never know where something interesting is going to come from. Johann Hari has written a book about the war on drugs- its origins and rationale, and was interviewed on the atheist podcast, Point of Inquiry. The story is remarkable and shocking.

Way back, when the Victorians such as Arthur Conan Doyle were doing cocaine and opium, no one thought to criminalize such drugs. If you wanted to kill yourself, go ahead. But temperance (vs alcohol) set the stage for the criminalization concept, (and its utter failure), in an extreme case of historical irony. As Hari portrays it, the end of prohibition led its leaders and bureaucratic apparatus to look for other ways to retain power and stay occupied. Presto.. the heightened criminalization of heroin and marijuana, which had begun in with the Harrison Narcotics act in 1914. Note that the most addictive drugs of all were left untouched and continue to kill millions of people yearly ... nicotine and alcohol.

There is no question that this is a class-based construction, and Hari cites intense racism as a motivating factor, as the "hard" drugs such as opium, heroin, and marijuana were thought to be favored by the lower classes. Billy Holiday is a big focus of story.

But the irony is that these addictions are not as deterministic as we have been led to believe. Rats as the model organism are brought in to show that while in the original experiments, they certainly preferred drug-laced water to plain water, these were run in bare cages where the rats were bored out of their minds, anxious, unhappy. If the same experiment is run in more normal conditions, in a physically and socially enriched environment, rats do not become addicted. They prefer a real, normal existence to one that is zonked out.

Hari also cites the experience of Switzerland, which has maintained a medical model of drug treatment (which we used to have, before it was taken over by the mania of the drug war). Addicts get their drugs prescribed, take them daily, and go about their lives. They also tend to quit on their own eventually, again preferring reality to a drugged life. It is a very low-stress solution to the problems of addiction. Much lower stress than the warfare that the US has used its leading post-war position to export around the world, severely damaging countries such as Colombia and Mexico in an effort to criminalize and stamp out what clearly can not be stamped out.

But the main issue is one of class and social support. When large swathes of the population are alienated, degraded, discarded, and dehumanized, drug addiction can naturally become a large-scale, scary problem. We would be tempted to treat it with zero tolerance, with mass incarceration, and a world-wide attempt to interdict the offending substances. Yet the problem lies not in the drugs, but in ourselves.

  • Organized crime, or organized religion?
  • Europe's little ice age may not have happened. But current warming is happening.
  • Not every employee at Citi was/is a criminal.
  • If corporations are people, they aren't very good people.
  • Market failure or government failure? The ideology of monetarism.
  • One reason why education will never solve inequality: " the continuing tragedy of adjunctification".
  • A glimmer of hope against the neonicotinoids.
  • Why does anyone take Putin's nuclear blackmail seriously?