Perception as controlled hallucination - predictive processing and the nature of conscious experience

I've now read several times through a fascinating Edge.org conversation with philosopher Andy Clark. I suggest you read the piece, and here pass on some edited clips. First, his comments on most current A.I. efforts:

There's something rather passive about the kinds of artificial intelligence ...[that are]...trained on an objective function. The AI tries to do a particular thing for which it might be exposed to an awful lot of data in trying to come up with ways to do this thing. But at the same time, it doesn't seem to inhabit bodies or inhabit worlds; it is solving problems in a disembodied, disworlded space. The nature of intelligence looks very different when we think of it as a rolling process that is embedded in bodies or embedded in worlds. Processes like that give rise to real understandings of a structured world.

Then, his ideas on how our internal and external worlds are a continuum:

Perception itself is a kind of controlled hallucination. You experience a structured world because you expect a structured world, and the sensory information here acts as feedback on your expectations. It allows you to often correct them and to refine them. But the heavy lifting seems to be being done by the expectations. Does that mean that perception is a controlled hallucination? I sometimes think it would be good to flip that and just think that hallucination is a kind of uncontrolled perception.

The Bayesian brain, predictive processing, hierarchical predictive coding are all, roughly speaking, names for the same picture in which experience is constructed at the shifting borderline between sensory evidence and top-down prediction or expectation. There's been a big literature out there on the perceptual side of things. It's a fairly solid literature. What predictive processing did that I found particularly interesting—and this is mostly down to a move that was made by Karl Friston—was apply the same story to action. In action, what we're doing is making a certain set of predictions about the shape of the sensory information that would result if I were to perform the action. Then you get rid of prediction errors relative to that predicted flow by making the action.

There's a pleasing symmetry there. Once you've got action on the table in these stories—the idea is that we bring action about by predicting sensory flows that are non actual and then getting rid of prediction errors relative to those sensory flows by bringing the action about—that means that epistemic action, as it's sometimes called, is right there on the table. Systems like that cannot just act in the world to fulfill their goals; they can also act in the world so as to get better information to fulfill their goals. And that's something that active animals do all the time. The chicken, when it bobs its head around, is moving its sensors around to get information that allows it to do depth perception that it can't do unless it bobs its head around...Epistemic action, and practical action, and perception, and understanding are now all rolled together in this nice package.

An upshot here is that there's no experience without the application of some model to try to sift what is worthwhile for a creature like you in the signal and what isn't worthwhile for a creature like you.

Apart from the exteroceptive signals that we take in from vision, sound, and so on, and apart from the proprioceptive signals from our body that are what we predict in order to move our body around, there's also all of the interoceptive signals that are coming from the heart and from the viscera, et cetera...being subtly inflected by interoception information is part of what makes our conscious experience of the world the kind of experience that it is. So, artificial systems without interoception could perceive their world in an exteroceptive way, they could act in their world, but they would be lacking what seems to me to be one important dimension of what it is to be a conscious human being in the world.

Implicit racial bias is preserved by historical roots of social environments.

Payne et al. note that geographic differences in implicit racial bias correlate with the number of slaves in those areas in 1860.

Significance

Geographic variation in implicit bias is associated with multiple racial disparities in life outcomes. We investigated the historical roots of geographical differences in implicit bias by comparing average levels of implicit bias with the number of slaves in those areas in 1860. Counties and states more dependent on slavery in 1860 displayed higher pro-White implicit bias today among White residents and less pro-White bias among Black residents. Mediation analyses suggest that historical oppression may be transmitted into contemporary biases through structural inequalities, including disparities in poverty and upward mobility. Given the importance of contextual factors, efforts to reduce unintended discrimination might focus on modifying social environments that cue implicit biases in the minds of individuals.

Abstract

Implicit racial bias remains widespread, even among individuals who explicitly reject prejudice. One reason for the persistence of implicit bias may be that it is maintained through structural and historical inequalities that change slowly. We investigated the historical persistence of implicit bias by comparing modern implicit bias with the proportion of the population enslaved in those counties in 1860. Counties and states more dependent on slavery before the Civil War displayed higher levels of pro-White implicit bias today among White residents and less pro-White bias among Black residents. These associations remained significant after controlling for explicit bias. The association between slave populations and implicit bias was partially explained by measures of structural inequalities. Our results support an interpretation of implicit bias as the cognitive residue of past and present structural inequalities.

Over 50-fold difference between individuals in circadian melatonin sensitivity to evening light.

Phillips et al. probe how our high sensitivity to artificial light after dusk perturbs the circadian rhythm of our sleep hormone melatonin:

Significance

Electric lighting has fundamentally altered how the human circadian clock synchronizes to the day/night cycle. Exposure to light after dusk is pervasive in the modern world. We examined group-level sensitivity of the circadian system to evening light and the degree to which sensitivity varies between individuals. We found that, on average, humans are highly sensitive to evening light. Specifically, 50% suppression of melatonin occurred at less than 30 lux, which is comparable to or lower than typical indoor lighting used at night, as well as light produced by electronic devices. Significantly, there was a greater than 50-fold difference in sensitivity to evening light across individuals. Interindividual differences in light sensitivity may explain differential vulnerability to circadian disruption and subsequent impact on human health.

Abstract

Before the invention of electric lighting, humans were primarily exposed to intense (>300 lux) or dim (less than 30 lux) environmental light—stimuli at extreme ends of the circadian system’s dose–response curve to light. Today, humans spend hours per day exposed to intermediate light intensities (30–300 lux), particularly in the evening. Interindividual differences in sensitivity to evening light in this intensity range could therefore represent a source of vulnerability to circadian disruption by modern lighting. We characterized individual-level dose–response curves to light-induced melatonin suppression using a within-subjects protocol. Fifty-five participants (aged 18–30) were exposed to a dim control (less than 1 lux) and a range of experimental light levels (10–2,000 lux for 5 h) in the evening. Melatonin suppression was determined for each light level, and the effective dose for 50% suppression (ED₅₀) was computed at individual and group levels. The group-level fitted ED₅₀ was 24.60 lux, indicating that the circadian system is highly sensitive to evening light at typical indoor levels. Light intensities of 10, 30, and 50 lux resulted in later apparent melatonin onsets by 22, 77, and 109 min, respectively. Individual-level ED₅₀ values ranged by over an order of magnitude (6 lux in the most sensitive individual, 350 lux in the least sensitive individual), with a 26% coefficient of variation. These findings demonstrate that the same evening-light environment is registered by the circadian system very differently between individuals. This interindividual variability may be an important factor for determining the circadian clock’s role in human health and disease.

Mechanism of exercise and antioxidant stimulation of memory and new nerve cell growth

On reading this article by Yook et al. I promptly ordered a bottle of 10 mg astaxanthin capsules to add to my normal array of supplements (and exercise).

Significance

Leptin (LEP, a small protein hormone), produced and acting in the hippocampus, mediates enhancement by mild exercise (ME) of hippocampus-related memory and neurogenesis, which are further increased by an antioxidant carotenoid, astaxanthin (AX). Both are facilitated by the administration of ME or AX alone. The up-regulation of the LEP gene and LEP protein expression in the hippocampus by ME is further elevated when combined with AX. Consistently, the combined interventions increased hippocampal LEP protein. In LEP-deficient ob/ob mice, LEP replacement in the brain restored the ability of ME+AX to enhance hippocampal function. Thus, a combined lifestyle intervention based on ME, including yoga and tai chi, and specific dietary supplements that include antioxidants may together improve cognition and possibly retard cognitive decline in humans.

Abstract

Regular exercise and dietary supplements with antioxidants each have the potential to improve cognitive function and attenuate cognitive decline, and, in some cases, they enhance each other. Our current results reveal that low-intensity exercise (mild exercise, ME) and the natural antioxidant carotenoid astaxanthin (AX) each have equivalent beneficial effects on hippocampal neurogenesis and memory function. We found that the enhancement by ME combined with AX in potentiating hippocampus-based plasticity and cognition is mediated by leptin (LEP) made and acting in the hippocampus. In assessing the combined effects upon wild-type (WT) mice undergoing ME with or without an AX diet for four weeks, we found that, when administrated alone, ME and AX separately enhanced neurogenesis and spatial memory, and when combined they were at least additive in their effects. DNA microarray and bioinformatics analyses revealed not only the up-regulation of an antioxidant gene, ABHD3, but also that the up-regulation of LEP gene expression in the hippocampus of WT mice with ME alone is further enhanced by AX. Together, they also increased hippocampal LEP (h-LEP) protein levels and enhanced spatial memory mediated through AKT/STAT3 signaling. AX treatment also has direct action on human neuroblastoma cell lines to increase cell viability associated with increased LEP expression. In LEP-deficient mice (ob/ob), chronic infusion of LEP into the lateral ventricles restored the synergy. Collectively, our findings suggest that not only h-LEP but also exogenous LEP mediates effects of ME on neural functions underlying memory, which is further enhanced by the antioxidant AX.

Enhancing longevity by removing deteriorated body cells.

Here I pass on both the introductory summary and concluding paragraph of van Deursen's review of efforts to enhance longevity by removing body cells that have deteriorated and become dysfunctional (SNCs).

The estimated “natural” life span of humans is ∼30 years, but improvements in working conditions, housing, sanitation, and medicine have extended this to ∼80 years in most developed countries. However, much of the population now experiences aging-associated tissue deterioration. Healthy aging is limited by a lack of natural selection, which favors genetic programs that confer fitness early in life to maximize reproductive output. There is no selection for whether these alterations have detrimental effects later in life. One such program is cellular senescence, whereby cells become unable to divide. Cellular senescence enhances reproductive success by blocking cancer cell proliferation, but it decreases the health of the old by littering tissues with dysfunctional senescent cells (SNCs). In mice, the selective elimination of SNCs (senolysis) extends median life span and prevents or attenuates age-associated diseases. This has inspired the development of targeted senolytic drugs to eliminate the SNCs that drive age-associated disease in humans.

As knowledge of the fundamental biology and vulnerabilities of SNCs expands, the rational design of targeted senolytics is expected to yield therapies to eliminate SNCs that drive degeneration and disease. This positive outlook is based on successes in oncology and because the main limitation of cancer therapies—the clonal expansion of drug-resistant cells—does not apply to SNCs. Additional confidence comes from the recent progress in bringing senolytic agents into clinical trials. The first clinical trial is testing UBX0101 for the treatment of osteoarthritis of the knee. Another drug, UBX1967, a BCL-2 family inhibitor specifically tailored for diseases of the aging eye, is also advancing to human testing. Multiple clinical trials treating diverse diseases of aging with senolytic drugs are expected to follow soon. This includes two-step cancer treatment approaches whereby malignant cells are first forced into a senescent state by one drug and then eliminated with a senolytic agent. Success in these first clinical studies is the next critical milestone on the road to the development of treatments that can extend healthy longevity in people.

Why can we read only one word at a time?

Fascinating work from White et al.:

Significance

Because your brain has limited processing capacity, you cannot comprehend the text on this page all at once. In fact, skilled readers cannot even recognize just two words at once. We measured how the visual areas of the brain respond to pairs of words while participants attended to one word or tried to divide attention between both. We discovered that a single word-selective region in left ventral occipitotemporal cortex processes both words in parallel. The parallel streams of information then converge at a bottleneck in an adjacent, more anterior word-selective region. This result reveals the functional significance of subdivisions within the brain’s reading circuitry and offers a compelling explanation for a profound limit on human perception.

Abstract

In most environments, the visual system is confronted with many relevant objects simultaneously. That is especially true during reading. However, behavioral data demonstrate that a serial bottleneck prevents recognition of more than one word at a time. We used fMRI to investigate how parallel spatial channels of visual processing converge into a serial bottleneck for word recognition. Participants viewed pairs of words presented simultaneously. We found that retinotopic cortex processed the two words in parallel spatial channels, one in each contralateral hemisphere. Responses were higher for attended than for ignored words but were not reduced when attention was divided. We then analyzed two word-selective regions along the occipitotemporal sulcus (OTS) of both hemispheres (subregions of the visual word form area, VWFA). Unlike retinotopic regions, each word-selective region responded to words on both sides of fixation. Nonetheless, a single region in the left hemisphere (posterior OTS) contained spatial channels for both hemifields that were independently modulated by selective attention. Thus, the left posterior VWFA supports parallel processing of multiple words. In contrast, activity in a more anterior word-selective region in the left hemisphere (mid OTS) was consistent with a single channel, showing (i) limited spatial selectivity, (ii) no effect of spatial attention on mean response amplitudes, and (iii) sensitivity to lexical properties of only one attended word. Therefore, the visual system can process two words in parallel up to a late stage in the ventral stream. The transition to a single channel is consistent with the observed bottleneck in behavior.

Why high-class people get away with incompetence.

Belmi et al. do four experiments suggesting that people who come from a higher social class are more likely to have an inflated sense of their skills — even when tests proved that they are average. This unmerited overconfidence is interpreted by strangers as competence. This highlights yet another way that family wealth and parents' education confers an advantage in getting ahead in life.

Understanding how socioeconomic inequalities perpetuate is a central concern among social and organizational psychologists. Drawing on a collection of findings suggesting that different social class contexts have powerful effects on people’s sense of self, we propose that social class shapes the beliefs that people hold about their abilities, and that this, in turn, has important implications for how status hierarchies perpetuate. We first hypothesize that compared with individuals with relatively low social class, individuals with relatively high social class are more overconfident. Then, drawing on research suggesting that overconfidence can confer social advantages, we further hypothesize that the overconfidence of higher class individuals can help perpetuate the existing class hierarchy: It can provide them a path to social advantage by making them appear more competent in the eyes of others. We test these ideas in four large studies with a combined sample of 152,661 individuals. Study 1, a large field study featuring small-business owners from Mexico, found evidence that individuals with relatively high social class are more overconfident compared with their lower-class counterparts. Study 2, a multiwave study in the United States, replicated this result and further shed light on the underlying mechanism: Individuals with relatively high (vs. low) social class tend to be more overconfident because they have a stronger desire to achieve high social rank. Study 3 replicated these findings in a high-powered, preregistered study and found that individuals with relatively high social class were more overconfident, even in a task in which they had no performance advantages. Study 4, a multiphase study that featured a mock job interview in the laboratory, found that compared with their lower-class counterparts, higher-class individuals were more overconfident; overconfidence, in turn, made them appear more competent and more likely to attain social rank.

Policy evaluation by randomized trials may provoke greater objections than implementing them untested.

Meyer et al. find that experiments comparing two unobjectionable policies or treatments can generate more objections than simply proceeding to implement them.

Significance

Randomized experiments—long the gold standard in medicine—are increasingly used throughout the social sciences and professions to evaluate business products and services, government programs, education and health policies, and global aid. We find robust evidence—across 16 studies of 5,873 participants from three populations spanning nine domains—that people often approve of untested policies or treatments (A or B) being universally implemented but disapprove of randomized experiments (A/B tests) to determine which of those policies or treatments is superior. This effect persists even when there is no reason to prefer A to B and even when recipients are treated unequally and randomly in all conditions (A, B, and A/B). This experimentation aversion may be an important barrier to evidence-based practice.

Abstract

Randomized experiments have enormous potential to improve human welfare in many domains, including healthcare, education, finance, and public policy. However, such “A/B tests” are often criticized on ethical grounds even as similar, untested interventions are implemented without objection. We find robust evidence across 16 studies of 5,873 participants from three diverse populations spanning nine domains—from healthcare to autonomous vehicle design to poverty reduction—that people frequently rate A/B tests designed to establish the comparative effectiveness of two policies or treatments as inappropriate even when universally implementing either A or B, untested, is seen as appropriate. This “A/B effect” is as strong among those with higher educational attainment and science literacy and among relevant professionals. It persists even when there is no reason to prefer A to B and even when recipients are treated unequally and randomly in all conditions (A, B, and A/B). Several remaining explanations for the effect—a belief that consent is required to impose a policy on half of a population but not on the entire population; an aversion to controlled but not to uncontrolled experiments; and a proxy form of the illusion of knowledge (according to which randomized evaluations are unnecessary because experts already do or should know “what works”)—appear to contribute to the effect, but none dominates or fully accounts for it. We conclude that rigorously evaluating policies or treatments via pragmatic randomized trials may provoke greater objection than simply implementing those same policies or treatments untested.

Is technology really reshaping our consciousness?

My first reaction on reading a recent Op-Ed piece by David Brooks "When Trolls and Crybullies Rule the Earth" was to say 'Yes!', and think I should immediately fire off some selected clips in a MindBlog post. I'm glad I waited a bit, because as I look at it again I feel he has erred on the side of being an alarmist drama queen to grab our attention. The article begins with:

Over the past several years, teenage suicide rates have spiked horrifically....What's going on?

Several sources show an increasing rate from 2010 to 2017, but a look at Centers for Disease Control data shows higher rates for the late 1980's and early 1990's. He continues:

My answer starts with technology but is really about the sort of consciousness online life induces.

Brooks then describes transformations of human consciousness as if they have replaced, rather than adding to and enhancing, older forms of consciousness:  the shift from an oral to a printed culture centuries ago and the current shift from printed to electronic communication.

Attention and affection have gone from being private bonds to being publicly traded goods...up until recently most of the attention a person received came from family and friends and was pretty stable. But now most of the attention a person receives can come from far and wide and is tremendously volatile...your online post can go viral and get massively admired or ridiculed, while other times your post can leave you alone and completely ignored. Communication itself, once mostly collaborative, is now often competitive, with bids for affection and attention. It is also more manipulative — gestures designed to generate a response.

But... were not the old fashioned kinds of attention exchanged personally or in crowds of people, rather than electronically, labile and competitive with constant bids for affection and attention? Electronics may have amplified what was happening, but it didn't fundamentally transform it. Vicious gossip can be exchanged in old fashioned personal or newer less personal electronic ways. Online platforms may be an amplifier of our traits, but they don't basically transform them. Trolls and crybullies have always been with us. Still, Brooks makes good points, even if a bit exaggerated:

The internet has become a place where people communicate out of their competitive ego: I’m more fabulous than you (a lot of Instagram). You’re dumber than me (much of Twitter). It’s not a place where people share from their hearts and souls.

Of course, people enmeshed in such a climate are more likely to feel depressed, to suffer from mental health problems. Of course, they are more likely to see human relationship through the abuser/victim frame, and to be acutely sensitive to any power imbalance. Imagine you’re 17 and people you barely know are saying nice or nasty things about your unformed self. It creates existential anxiety and hence fanaticism.

The wisdom of partisan crowds.

Fascinating and counterintuitive findings from Becker et al., who find that the wisdom of crowds is robust to partisan bias:

Significance

Normative theories of deliberative democracy are based on the premise that social information processing can improve group beliefs. Research on the “wisdom of crowds” has found that information exchange can increase belief accuracy in many cases, but theories of political polarization imply that groups will become more extreme—and less accurate—when beliefs are motivated by partisan political bias. While this risk is not expected to emerge in politically heterogeneous networks, homogeneous social networks are expected to amplify partisan bias when people communicate only with members of their own political party. However, we find that the wisdom of crowds is robust to partisan bias. Social influence not only increases accuracy but also decreases polarization without between-group network ties.

Abstract

Theories in favor of deliberative democracy are based on the premise that social information processing can improve group beliefs. While research on the “wisdom of crowds” has found that information exchange can increase belief accuracy on noncontroversial factual matters, theories of political polarization imply that groups will become more extreme—and less accurate—when beliefs are motivated by partisan political bias. A primary concern is that partisan biases are associated not only with more extreme beliefs, but also with a diminished response to social information. While bipartisan networks containing both Democrats and Republicans are expected to promote accurate belief formation, politically homogeneous networks are expected to amplify partisan bias and reduce belief accuracy. To test whether the wisdom of crowds is robust to partisan bias, we conducted two web-based experiments in which individuals answered factual questions known to elicit partisan bias before and after observing the estimates of peers in a politically homogeneous social network. In contrast to polarization theories, we found that social information exchange in homogeneous networks not only increased accuracy but also reduced polarization. Our results help generalize collective intelligence research to political domains.

Inequality brokered

Public opinion about sexual minorities has improved dramatically in recent years, but Sun and Gao do an analysis showing that discriminatory behavior in lending practices has been ongoing and gone unchecked.

Significance

We propose a method to infer people’s sexual orientation indirectly through gender disclosure of the borrower and coborrower in a mortgage. Furthermore, we examine lending practices toward same-sex borrowers and its spillover effects. We attempt to extend the research on race/gender discrimination by systematically investigating the potentially different lending treatment toward same-sex borrowers. The data reveal that, compared with otherwise similar different-sex applicants, same-sex applicants are 73.12% more likely to be denied, and they tend to be charged up to 0.2% higher fees/interest. Furthermore, neighborhoods’ higher same-sex population density adversely affects both same-sex and different-sex borrowers’ lending experiences. Our method might approximately measure the US homosexual population distribution up to the census tract level annually over decades.

Abstract

Using massive US mortgage lending data, we propose a method to infer a borrower’s sexual orientation indirectly without a self-identification requirement and demonstrate the method’s potential to approximately measure the sexual orientation of the US population at the local level annually over decades. We continue to examine the lending practices to same-sex borrowers and its spillover effects. The persistent results since 1990 reveal that, in contrast with otherwise comparable different-sex loan applicants, the approval rate for same-sex applicants is ∼3–8% lower. Furthermore, conditional on approval, lenders, on average, charge about 0.02–0.2% higher interest to same-sex borrowers, which is equivalent to an annual total of $8.6 million to $86 million in additional interest/fees nationwide. Meanwhile, we find that same-sex borrowers are less risky overall, as they exhibit similar default risk but lower prepayment risk. Finally, we document findings of spillover effects. That is, when the share of a neighborhood’s same-sex population increases, both same-sex and different-sex borrowers seem to experience more unfavorable lending outcomes overall. The findings should raise enough concerns to warrant further investigations.

Blueprint - Nicholas Christakis' new book on evolutionary origins of a good society

This opinion piece by Frank Bruni in the NYTimes motivated me to download and read Nicholas Christakis' Magnum Opus “Blueprint” (very much in the 'everything you need to know about humans' spirit of Sapolsky's "Behave" and Harari's "Sapiens," and "Homo Deus," and "21 Lessons," all books that I have made the subject of previous posts.). It echoes Pinker's emphasis on the more positive aspects of human nature and progress. It is a very engaging read, and not amenable to a simple summary, but here is a bit from his introduction:

How can people be so different from—even go to war with—one another and yet also be so similar? The fundamental reason is that we each carry within us an evolutionary blueprint for making a good society.

Genes do amazing things inside our bodies, but even more amazing to me is what they do outside of them. Genes affect not only the structure and function of our bodies; not only the structure and function of our minds and, hence, our behaviors; but also the structure and function of our societies. This is what we recognize when we look at people around the world. This is the source of our common humanity.

Natural selection has shaped our lives as social animals, guiding the evolution of what I call a “social suite” of features priming our capacity for love, friendship, cooperation, learning, and even our ability to recognize the uniqueness of other individuals. Despite all the trappings and artifacts of modern invention—our tools, agriculture, cities, nations—we carry within us innate proclivities that reflect our natural social state, a state that is, as it turns out, primarily good, practically and even morally. Humans can no more make a society that is inconsistent with these positive urges than ants can suddenly make beehives.

I believe that we come to this sort of goodness just as naturally as we come to our bloodier inclinations. We cannot help it. We feel great when we help others. Our good deeds are not just the products of Enlightenment values. They have a deeper and prehistoric origin. The ancient tendencies that form the social suite work together to bind communities, specify their boundaries, identify their members, and allow people to achieve individual and collective objectives while at the same time minimizing hatred and violence. For too long, in my opinion, the scientific community has been overly focused on the dark side of our biological heritage: our capacity for tribalism, violence, selfishness, and cruelty. The bright side has been denied the attention it deserves.