The Limits of Reason: Cognitive Psychology, The Epistemological Crisis, and Epistemic Humility, #4.

(Palazzi, 2023)

TABLE OF CONTENTS

1. Introduction

2. Background: The Cognitive Limits of Rationality

3. Cognitive Blindspots

4. The Myth of the All-Seeing Eye: The Limits of Perception

5. The Epistemological Crises

6. Conclusion

The essay that follows is being published in six installments, one per section; this is the fourth installment.

But you can also download and read or share a .pdf of the complete text of the essay, including the REFERENCES, by scrolling to the bottom of this post and clicking on the Download tab.

The Limits of Reason: Cognitive Psychology, The Epistemological Crisis, and Epistemic Humility, #4

There is a crack, a crack in everything

That’s how the light gets in. (Cohen, 1992)

[H]ow would we feel if science came up against experimental and intellectual brick walls, so that after centuries of trying, man finally concluded that the world was constructed – if upon intelligible principles at all – upon principles so bizarre as to be perfectly undiscoverable or unfathomable by the human mind? What if [humankind] became totally convinced that the world simply could not be understood, that the world is and always must remain an intellectual surd? Science might then continue at it pertains to technology, but not as it pertains to theory. What if all hope of theoretical understanding were permanently lost? (Davis, 1987: 293)

Only those who stop at the right moment prosper in philosophy, those who accept the limit and the comfort of a reasonable level of worry. Every problem, if one touches the bottom, leads to bankruptcy and leaves the intellect naked: No more questions and no more answers in a space without horizons. The questions turn against the mind which conceived them: It becomes their victim. Everything becomes hostile: [their] own solitude, [their] own audacity, absolute opacity, and the manifest nothingness. Woe to [that person] who, having reached a certain point of the essential, has not stopped! History shows that the thinkers who climbed to the limit of the ladder of questions, who laid their foot on the last rung, on that of the absurd, have given to posterity an example of sterility, whereas their peers, who stopped half-way, have fertilized the mind’s flow; they have been useful to their fellows, they have passed down some well-crafted idol, a few polished superstitions, a few errors dressed up as principles, and a system of hopes. (Cioran, 1949: pp. 115-116)

4. The Myth of the All-Seeing Eye: The Limits of Perception

It is also worth noting that the position of the leading critics of the unbounded rationality position, Herbert Simon (Simon, 1979) and Daniel Kahneman (Kahneman, 2011), have come in for criticism for containing metaphysical and methodological biases of their own, namely for accepting an idea of an “all-seeing eye” (Hoffman & Prakash, 2014; Koenderink, 2014; Felin et al., 2017). Felin et al. make the criticism that while rejecting the idea of unbounded rationality of agents (i.e. perfect information, no uncertainty, and optimal decision-making), the Simon-Kahneman school have replaced “economic omniscience” with “perceptual omniscience, with the metaphor of an ‘all-seeing eye’” (Felin et al., 2017: p. 1040), a type of God’s eye view of the universe, embodied in the thesis of metaphysical realism. Metaphysical realism, roughly stated, says that objects of the world exist independent of thought/cognition, and their natures are ontologically independent of conceptions of them (Khlentzos, 2021). The critique of the all-seeing eye metaphor can be seen as a close philosophical relative of Richard Rorty’s earlier argument in Philosophy and the Mirror of Nature, that the attempt to find some sort of essence to reality (“our glassy essence”), whereby language mirrors reality via a representational theory of perception and a correspondence theory of truth, is inherently flawed (Rorty, 1979).

The “all-seeing eye” metaphor, Felin et al. believe, runs through a large number of theories in the social sciences, as well as theories of cognition, including: (1) Bayesian models of rationality and cognition, (2) various approaches to decision-making and classical decision theory, (3) a number of philosophies of mind, (4) rational expectation theory in neo-classical economics, (5) ideal versus naïve observer analysis, (6) adaptive control and cognitive architecture theories, and (7) various models of optimal foraging and general models of “computational rationality and intelligence” (Felin, et al., 2017: p. 1041). The all-seeing eye metaphor is manifest in various ways in many contemporary theories of rationality, for example, in assertions about global rationality of some systems of thought, or

in the form of a scientist who imputes illusion, bias, or other forms of error or veridically to subjects—when they fall short of omniscience. (Felin et al., 2017, 1041)

Theories of rationality typically assume that someone, such as the knowing subject, the scientists, or more abstractly, “the system as a whole,” is capable, through veridical perception, to ascertain the objective facts of the matter, and determine the best actions for an agent to take, in order to achieve various ends (Felin et al., 2017: p. 1041).

Nevertheless, there are a number of objections to the replacement of economic omniscience with perceptual omniscience, as described by Felin et al. Perception does not map “truth” in the way of “ideal observer” theories of perception, based on a hypothetical observer who has optimal perception on a set task. For humans, there is a closer match between perceptual performance and practical utility, linked to evolutionary fitness and survival, than between perceptual performance and acquiring so-called objective truths about “reality” (Felin et al., 2017: p. 1043). Organisms, including humans, do not exist in an objective perceptual environment; instead, perception is conditioned by what they are: perception is organism-specific and the nature of the organism determines what is perceived, and what can be perceived (Felin et al., 2017: p. 1043).

More fundamentally, perception does not involve a world-to-mind mapping in a camera-like fashion of representation of the “true” external world to “true” internal conceptions of the world, which is a common foundational assumption of the cognitive sciences (Koenderink et al., 2014). Felin et al., note that the visual illusions, such as the Ponzo illusion, have been interpreted by cognitive psychologists as showing the fallibility, limits, and biases of human perception (Gregory, 2005), and are “an artefact of the problem of singularity and exhaustively representing objective reality in the first place” (Felin et al., 2017: p. 1046).

Indeed, some visual illusions show not merely bias in human perception, but present a “reality” which is scientifically false, if not impossible, such as the seeming incompatibility between the observable world and the world of quantum mechanics, as presented in the “two tables” problem of Eddington, where the table of physics is mostly empty space, and what is not empty space, is filled with particles that have properties that large-scale objects do not have (Eddington, 1927; Bub, 1999).

The neural mechanics involved in a number of optical illusions, such as the “hypnotic vibes,” various patterns that fool the brain into perceiving motion, are not yet known (Sarcone, 2013). Likewise, for the even more philosophically and mathematically interesting perception of impossible objects, and the perception of motion as inconsistent (Mortensen, 2014). For example, impossible images were devised by Oscar Reutersvärd (1915-2002), M.C. Escher (1898-1972) and Roger Penrose (1931-). A famous example is Escher’s lithograph print Relativity (1953), which depicts a world in which not only does the law of gravity not hold, but also there’s an impossible situation whereby moving up the stairway leads to moving down the stairway, simultaneously. This impossibility is also depicted in the lithographs, Ascending and Descending (1960), and Waterfall (1961). The latter depicts a perpetual motion machine, as water flowing down, also flows simultaneously up as well. These impossible pictures have been analysed within a framework of a paraconsistent geometry (Mortensen, 2010).

It is common enough for cognitive psychologists to conclude from the study of illusions and the fallibility of the human perceptual system, that human perception does not operate as a video reproduction of reality, but rather is an interpretative process influenced by a range of factors such as prior beliefs and knowledge, experience and expectations even with respect to simple perceptual properties such as color, shape, and size (Gregory & Heard, 1979; Pronin et al., 2002). However, the more interesting philosophical thesis has been put forward by Brian Rogers, namely, that we are deluded about the nature of illusions because there is no epistemologically satisfactory way of distinguishing between perceptual experiences regarded as veridical, and those regarded as illusions (Rogers, 2014). The problem is that illusions are widely regarded as “departures from reality” (Gregory, 2009: p. 9), but we do not know what reality is, outside of the working of our perceptual system, which raises the classical epistemological skepticism problem of the justification of the existence of the external world. If there is no single objective reality “out there” by which perceptions can be compared to in some pre-theoretical way, and hence no “all seeing eye,” as metaphysical realists suppose, and no way the world actually is (Koenderink, 2014), then how can anything at all exist beyond sensory perceptions? As Koenderink puts it:

The very notion of veridicality itself, so often invoked in vision studies, is void. Strictly speaking, veridicality applies to the description of an external observer (Watcher, say) who watches both the subject–agent and its environment. The Watcher has to approximate the All Seeing Eye sufficiently for the purposes of the experiment. This implies that the Watcher knows more of the environment than the agent possibly can. This often implies pointer readings: for instance, electrical measurements in the study of electroreception in sharks, caliper gauges in the study of human acuity, and so forth. Then perception may sometimes be called ‘veridical’ relative to the knowledge of the Watcher. This is a very tricky business, because human Watchers lack the All Seeing Eye. They too are only directly aware of their user interfaces—even when using instruments. This offers interesting opportunities for infinite regress. Who Watches the Watcher? Only Big Brother has the All Seeing Eye. (Koenderink, 2014, 5)

However, one of the consequences of this anti-realist turn is precisely noted by Rogers:

the distinction between the veridical and the illusory becomes meaningless and we are forced to regard either all our perceptions as illusions or none of them, which is hardly helpful or informative. (Rogers, 2014: p. 844)

This problem confronts other cognitive psychologists such as Donald Hoffman (Hoffman, 2019a, 2019b), who explicitly argues that human perception is non-veridical and did not evolve to reveal the truth about reality, and that what we deal with is an interface with reality. This is the Interface Theory of Perception (ITP). Hoffman compares our senses to the desktop interface on say a laptop computer, where the interface does not reveal the hidden truth about the inner electronics of the system, but enables tasks to be completed. Likewise, for humans, evolution has shaped human senses so that there can be an interaction with reality to preserve fitness and survival. According to Hoffman “fitness beats truth,” with the perceptual system being designed for fitness, not truth. As he says:

Spacetime is your 3-D desktop. It is not the ancient stage for a reality play in which we are recent bit players. You create spacetime when you look. You are the scenic designer that creates spacetime, stars, planets, mountains, and oceans with a glance, and then erases them with a blink. There is a reality that exists even if you do not look, but it is unlike the spacetime and objects that you create when you interact with reality.

Why is it that when I see a bus, others usually do too? Because, as members of a species, we have a similar interface. Why is it that the bus can kill me, even if I do not see it? Because there is a reality that is objective, that exists even if I do not perceive it. That reality can affect me, whether I perceive it or not. But that reality has no buses in it. Buses are my interface icons, created as I interact with that reality. (Hoffman, 2019a: p. 66)

The problem here, which Hoffman has addressed, is that this position is arguably self-undermining and that he self-implicates his own core evidence, namely, that from evolutionary theory. As Hoffman says in an interview with David Gruber:

Now, the rejoinder then is, well, evolution by natural selection, as standardly formulated, assumes that physical objects like DNA exist and have definite properties. And I’m saying that even space itself doesn’t exist. Space itself is just a data structure that we create if our senses evolved. And so, why is it the case that I’m not refuting myself? I’ve used evolution to prove that evolution is false. And it turns out, I’ve used — when I use evolutionary game theory, I don’t use all of the evolutionary theory. I used what’s called “the algorithmic core,” what Dennett and Dawkins call Universal Darwinism.… Variation, collection, retention — that program which is at the heart of evolution. And evolutionary game theory is the mathematics that captures that heart of evolutionary theory. Now, evolutionary game theory makes no ontological assumptions. It doesn’t assume anything about space and time and matter and so forth. It’s an algorithm. It says anything that can vary and have retention and selection is subject to evolution. And so, Dennett and Dawkins, for example, are happy to talk about memes, ideas that evolve, and scientific theories themselves as all subject to evolution, right? So what I have done in my theorem with Chetan Prakash, my collaborator who proved the theorem, and my graduate students, Justin Mark and Brian Marion—what we found is that the algorithmic core of evolution by natural selection is incompatible with the side assumptions that are made in standard evolutionary biology; namely, that physical space exists and that physical organisms with physical DNA exist. These are all symbols that we’re using, pointing to a deeper reality. (Gruber, 2021: pp. 174-175)

The “deeper reality” will consist of a mathematically precise dynamic system of conscious agents, which are fundamental.

Now, that could well be so, and there is frontier philosophical and metaphysical work by neo-Kantian philosopher Robert Hanna which could have helped Hoffman, if Hoffman had taken the road of metaphysics, and transcendental idealism, rather than evolutionary algorithms, however trendy that may be (Hanna & Maiese, 2009; Hanna, 2021, 2022a). Given Hoffman’s own philosophical framework, evolutionary arguments, based upon an “ algorithmic core,” allegedly show that evolution by natural selection is incompatible with the assumption that physical space exists and that physical organisms will exist. But that cannot be right, since whatever evolution is, it involves at a minimum a change in material things, such as organisms. If not, what could possibly be the subject of evolution? A mere abstract mathematical core will not deliver the world he wants. And, beyond that, if the existence of space-time and material objects such as organisms can be called into question by Hoffman, why should we assume that evolution exists as well? Why suppose that algorithms exist? Perhaps whatever there is, was created by God, or an evil demon, or even a computer simulation by incomprehensive cosmic computer programmers? Perhaps nothing really exists at all? Less exotically, from the perspective of evolution, consciousness itself is still in need of explanation, since if what counts are survival strategies, then why should consciousness exist, since organisms, if they existed could do just as well without it (Chalmers, 1995: p. 202)?

One possible reply to this criticism is to argue that this sort of circularity faces most foundational questions. The logocentric predicament, as defined by Harry Sheffer is that in “order to give an account of logic, we must presuppose and employ logic” (Sheffer, 1926: p. 228). Hanna has generalized the logocentric predicament to apply more widely to rationality, and philosophical systems building, to the effect that every attempt to either justify or criticize rationality presupposes rationality, which he calls the ratiocentric predicament (Hanna, 2006, 2023). Be that as it may, at least as regards Hoffman’s system, positing evolutionary theory, as he views it, is not a fundamental aspect of reality, in the same sense that logic and reason are, because we can conceive of a world in which evolutionary forces are only one factor among many others, or do not exist at all, or are simply different in nature to Hoffman’s conception of them.

This does restate one of the classical skeptical arguments against the existence of the external world, since if all that is present to us are perceptual representations, we seemingly lack any rationally justified belief that the cause of those representations is an external world, or even whether there is a cause at all (Slote, 1970). But, if this is so, then why accept the information that was used to establish this argument in the first place, for there seems to be an implicit realism that creeps into the anti-realism argument even though it ultimately undermines itself? What is the epistemic status of the perceptual data used to get this argument off the ground, and if it is merely “relative” too, then can we even trust the argument from relativity of perception (Smith, 1985)? In other words, cognitive psychology generates a “limit paradox.”

Eric Dietrich and Chris Fields hold that science itself generates these type of limit paradoxes, and their discussion, among other things, deals with the metaphysical challenges that quantum mechanics poses (Dietrich & Fields, 2015). For example, mainstream science presupposes that boundaries between systems can in principle be made, that the observer and the observed are separate entities, in the sense that at the very least, they’re spatially distinct. But quantum theory through quantum entanglement, challenges this mainstream scientific assumption:

By introducing entanglement as an inevitable physical consequence of dynamical interactions, quantum theory forecloses this possibility: a system could be objectively entangled with all other systems—… with all systems from the point of view of any competent observer—and hence objectively bounded only if it was isolated outside of the universe, a situation inconsistent with the standard definition of “the universe” as “everything,” as well as the assumption that quantum theory is complete. Hence quantum theory disallows the very assumptions that make the idealization of fully public, repeatable observations possible. (Dietrich & Fields, 2015)

The core assumption here, almost universally accepted, is that reproducibility is the hallmark of scientificity, a thesis Hanna has challenged (Hanna, 2023). Nevertheless, other examples could be given, so Dietrich-&-Fields’s conclusion is relevant by way of illustration:

If science is possible, it eventually produces results that undermine its assumptions and methods. … Hence science is impossible. But the result is in our possession, and science is necessary for our knowledge of the result. (Dietrich & Fields, 2015).

They conclude that these types of limit paradoxes (Priest, 2002) indicate a cognitive limit to science: “Our science is telling us, with increasing urgency … that the universe is not fully open to our comprehension” (Dietrich & Fields, 2015). And, they conclude, a key assumption of the Enlightenment was that human beings could obtain a God-like understanding of the universe, but this assumption is “overly optimistic,” and should be abandoned (Dietrich & Fields, 2015), a conclusion which can be further supported by a consideration of the epistemological crises.

Download