LessWrong Recommendations for Philosophy
#1
I wanted to initiate a discussion about this post on LessWrong:

http://lesswrong.com/lw/frp/train_philos...neman_not/

Quote:more Bayesian rationality, heuristics and biases, & debiasing, less informal "critical thinking skills";
more mathematical logic & theory of computation, less term logic;
more probability theory & Bayesian scientific method, less pre-1980 philosophy of science;
more psychology of concepts & machine learning, less conceptual analysis;
more formal epistemology & computational epistemology, less pre-1980 epistemology;
more physics & cosmology, less pre-1980 metaphysics;
more psychology of choice, less philosophy of free will;
more moral psychology, decision theory, and game theory, less intuitionist moral philosophy;
more cognitive psychology & cognitive neuroscience, less pre-1980 philosophy of mind;
more linguistics & psycholinguistics, less pre-1980 philosophy of language;
more neuroaesthetics, less aesthetics;
more causal models & psychology of causal perception, less pre-1980 theories of causation.
...

Author recommends this reading list:

Quote:Stanovich, Rationality and the Reflective Mind (2010)
Hinman, Fundamentals of Mathematical Logic (2005)
Russell & Norvig, Artificial Intelligence: A Modern Approach (3rd edition, 2009) — contains chapters which briefly introduce probability theory, probabilistic graphical models, computational decision theory and game theory, knowledge representation, machine learning, computational epistemology, and other useful subjects
Sipser, Introduction to the Theory of Computation (3rd edition, 2012) — relevant to lots of philosophical problems, as discussed in Aaronson (2011)
Howson & Urbach, Scientific Reasoning: The Bayesian Approach (3rd edition, 2005)
Holyoak & Morrison (eds.), The Oxford Handbook of Thinking and Reasoning (2012) — contains chapters which briefly introduce the psychology of knowledge representation, concepts, categories, causal learning, explanation, argument, decision making, judgment heuristics, moral judgment, behavioral game theory, problem solving, creativity, and other useful subjects
Dolan & Sharot (eds.), Neuroscience of Preference and Choice (2011)
Krane, Modern Physics (3rd edition, 2012) — includes a brief introduction to cosmology

Does anyone have opinions about these books?

I have acquired Pearl's book on Causality, and have read the first 25% of Kahneman's Thinking Slow and Fast. Thinking Slow and Fast is incredibly knowledgeable, and surely represents a much more scientific examination of 'human nature' than say, Freud. Other books in behavioral economics (e.g. Ariely's 'Predictably Irrational' also touch upon some of the topics. Kahneman's book is a thorough review of literature in behavioral economics. Kahneman's book has been discussed elsewhere on this forum: 1, 2, 3, 4


And for those with access to Amazon, I have created a convenient list of these books (created it as a wish-list, as I don't know any other way of creating lists in Amazon) http://www.amazon.com/registry/wishlist/...go_o_C-3_d
[+] 2 users Like karatalaamalaka's post
Reply
#2
That's a potentially very rewarding reading list.

However, some caveating maybe in order for one of the priorities suggested, namely
Quote: more neuroaesthetics, less aesthetics;

An anxiety with reducing aesthetics to neuroaesthetics may arise not exactly from inveterate preference for 'holism' over 'reductionism' as neuroaesthetic cheerleaders may hasten to assume, but of concerns at least two levels:

(i) The claims of certain experimental findings representing aesthetic universals, are undermined by the often inadequately diverse subject pools from which these findings are obtained. While this concern is more commonly expressed with behavioral economics claims which maybe ignoring important cultural influences while assuredly holding forth on human nature, the risk of universalizing ethical or aesthetic judgments of subjects from WEIRD societies does not go away by simply slapping on the prefix neuro- to these disciplines. Therefore, the caveats applicable to any Evolutionary Pscyhology claims must be contended with by neuro-aesthetic claims too.

(ii) Even if in theory there are indeed aesthetic universals, it would not mean immediately that aesthetic particulars cease to matter in aesthetic experience. Dr. V S Ramachandran who proposes a project of seeking laws of aesthetics, nevertheless in lectures like this one calls for 'neuro-aesthetics' to complement Art History, rather than subsume or obviate it. Development of any theory of aesthetics would involve both systematizing and empathizing influences, as well as the relevance of both universalizing and plularistic standpoints, as is the case in historical efforts at developing a theory of ethics. The intersubjective nature of aesthetic judgment would preclude a complete account of aesthetics by study of individual brains. Applying both the 'linguistic analogy' and the 'taste analogy' to the study of aesthetics, we may say that discerning the 'figural primitives of perceptual grammar' doesn't constitute the whole of aesthetics and the diversity of preferences remains a legitimate topic of study.

Since pluralism in artistic expression is a cultural asset worth conscientiously guarding, lending premature credence to misapprehended 'aesthetic universals' and allowing them unwittingly to homogenize standards of artistic appreciation is something for both art-lovers and science-lovers to be wary of.
Reply
#3
I have little to add about neuro-aesthetics, other than saying that I attended a talk by Vilayanur Ramachandran on neuro-aesthetics as it pertains to Indian art. The basic thesis of his talk was that Indian and other 'exotic' art was considered inferior by Europeans who were then hungover on realism. He explained that, neuro-aesthetically speaking, realism is in fact inferior to other art forms rich in abstract symbology.

Here are my first impressions of Pearl's 'Causality' [1]:

0. Probabilistic reasoning is important. We should all internalize the ability to interpret and model real-life events by using principles from probability theory.

1. Here is my non-technical attempt to describe what the book is about:
How do we formalize causality so that it can be included in models of various phenomena? Basic probabilistic reasoning does not take into account causal relationships. Tools such as correlation can only help us establish associative relationships. Formally, the distinction between associative and causal probabilistic models is this, to quote Pearl: "An associational concept is any relationship that can be defined in terms of a joint
distribution of observed variables, and a causal concept is any relationship that
cannot be defined from the distribution alone." [2]

Pearl paraphrases the standard fallacy of 'correlation does not imply causation' as: "one cannot substantiate causal
claims from associations alone, even at the population level—behind every causal conclusion there must lie some causal assumption that is not testable in observational studies." [2]

Historically, the first attempt to model causal relationships was by Sewell in 1922, where he proposed using arrows along with equal signs to indicate causality in equations that modeled physical phenomena. Such an approach is highly limited in its applicability.

The first requirement for modeling causality is to cast quantities in terms of conditional probabilities. But conditional probabilities in themselves aren't sufficient. We need to relate different quantities in terms of causation. This is achieved by utilizing graphs. By putting quantities at nodes of graphs, one can model causative relations by directional edges connecting these nodes. I will outline this in more detail in subsequent posts. I intended this post to just give a taste of what's in this book and what underlies causal inference (and the field of probabilistic graphical models, which I have familiarity with- more on that later)

2. Judea Pearl's book is intended for a technical audience. At minimum, some amount of post secondary mathematics is required to understand this book. Basic probability theory, algebra, and basic calculus are all strict prerequisites. A basic knowledge of graphs and networks will help you read the book faster.


----
References:

[1] Pearl, Judea. Causality: models, reasoning and inference. Vol. 29. Cambridge: MIT press, 2000.
[2] Pearl, Judea. "Causal inference in statistics: An overview." Statistics Surveys 3 (2009): 96-146.
[+] 2 users Like karatalaamalaka's post
Reply
#4
Quote:more probability theory & Bayesian scientific method, less pre-1980 philosophy of science;
more formal epistemology & computational epistemology, less pre-1980 epistemology;
more cognitive psychology & cognitive neuroscience, less pre-1980 philosophy of mind;
more linguistics & psycholinguistics, less pre-1980 philosophy of language;
more causal models & psychology of causal perception, less pre-1980 theories of causation.

On a closer look at the way Less Wrong reading recommendations are motivated and presented, something that is striking is the treatment of the 1980s as a definitive epoch in the history of science, to the extent of justifying a purge of all classical studies from the reading lists of students of philosophy. While the reading list by itself maybe valuable, the larger policy shift it seems to recommend seems more questionable.

It is true that the so-called Neural Winter, brought about in large part by Marvin Minsky's 1969 paper undermining the confidence and adoption of perceptron-based neural networks, began to thaw in the 1980s. The 1980s in a sense were indeed a 'Now is the winter of our discontent made glorious summer' hour for those sharing the interests of the Less Wrong authors, because it meant a fillip both for neural networks and Bayesian approaches (whose equivalence of sorts for many practical purposes was demonstrated later).

There is no denying that the wide adoption of Bayesian methods since the 1980s did occasion changes in the way business was done in fields like sensory neuroscience and linguistics. However, the underlying assumptions of the probabilistic models on which the Bayesian methods operate (for inference or parameter estimation) are assumptions that come from domain knowledge, influenced considerably by the history of the specific disciplines. It goes without saying that Bayesian inference is only a method to compare or evaluate hypotheses and not generate them! Therefore, treating the adoption of Bayesian methods as itself a fundamental revolution in a discipline is not tenable.

An example of a shift in the 'fundamental assumptions' or 'first principles' of a discipline, was one effected in large part by Noam Chomsky's idea of 'poverty of the stimulus' in linguistics. The argument briefly was that language competence is acquired by children using a surprisingly small amount of presented examples of correct language use, whereas a strict (Skinnerian) behaviorist mechanism of language acquisition would have demanded more copious inputs, thus suggesting the operation of innate mechanisms prior to experience. The identification and elucidation of such mechanisms, which constitutes much of the endeavor of the Cognitive Sciences, was occasioned in large part by this philosophical insight prior to the 1980s. Tellingly, Chomsky called it Plato's Problem, perhaps to the chagrin of the author of the Less Wrong recommendations who seems convinced that Plato is a distraction to students of science.

By treating methodological developments since the 1980s as fundamental overhauls of the disciplines themselves, the Less Wrong authors maybe making with regard to the history of science the mistake which that other evangelist of probabilistic reasoning Nicholas Nassim Taleb describes as common here: 'Experts and amateurs alike tend to overvalue the newest piece of information.' Indeed, the AI winter is itself an example of how the demise of certain fields that were prematurely written off, may have been greatly exaggerated. Most claims of having 'obviated philosophy' have also likewise been shown to be quite exasperated. Therefore, while the reading list is a useful one, it seems less recommended to buy into the mild end-of-history delusion about entire disciplines with which it seems to be sold.
[+] 1 user Likes arvindiyer's post
Reply
#5
I don't think LessWrong community seeks to obviate philosophy.

They seek to express ideas in the more exact (paradoxically :P) language of Bayesian statistics.

I liked this nice quote from Steven Pinker's most recent article, "Science is not your enemy"

Quote:These thinkers—Descartes, Spinoza, Hobbes, Locke, Hume, Rousseau, Leibniz, Kant, Smith—are all the more remarkable for having crafted their ideas in the absence of formal theory and empirical data. The mathematical theories of information, computation, and games had yet to be invented. The words “neuron,” “hormone,” and “gene” meant nothing to them. When reading these thinkers, I often long to travel back in time and offer them some bit of twenty-first-century freshman science that would fill a gap in their arguments or guide them around a stumbling block. What would these Fausts have given for such knowledge? What could they have done with it?


So really, a meta description:

We should seek to update a posterior, the classics of the Enlightenment using the language of Bayesian posteriors. :P
Reply
#6
(09-Aug-2013, 07:18 AM)karatalaamalaka Wrote: We should seek to update a posterior, the classics of the Enlightenment using the language of Bayesian posteriors. :P

...much as the Enlightenment itself was an update, a resetting of uncertainties as it were, of the Axial Age? Smartass The 'less classics - more Bayes' refrain in the Less Wrong list may be, well, less wrong in a readership where awareness of the Enlightenment is a given which of course is not something we can assume for readerships elsewhere. Therefore the recommendations on what to read (or what to read more of)
maybe less wrong than the accompanying recommendations on what not to read (or what to read less of).

(09-Aug-2013, 07:18 AM)karatalaamalaka Wrote: I liked this nice quote from Steven Pinker's most recent article, "Science is not your enemy"

Quote:These thinkers—Descartes, Spinoza, Hobbes, Locke, Hume, Rousseau, Leibniz, Kant, Smith—are all the more remarkable for having crafted their ideas in the absence of formal theory and empirical data. The mathematical theories of information, computation, and games had yet to be invented. The words “neuron,” “hormone,” and “gene” meant nothing to them. When reading these thinkers, I often long to travel back in time and offer them some bit of twenty-first-century freshman science that would fill a gap in their arguments or guide them around a stumbling block. What would these Fausts have given for such knowledge? What could they have done with it?

What Pinker suggests above is an interesting thought experiment of the sort: "How about we invite Adam Smith and Laurie Santos to the same panel?" and could serve as a useful trigger for a creative writing exercise. There are other questions one may ask as well, about the thought process, or creative process if we may, of thinkers who crafted their ideas in the absence (or near-absence or at any rate, paucity) of formal theory and empirical data. Many of Darwin's contemporaries probably had more copious and better-organized collections of naturalistic specimens than him, who infamously had forgotten to tag his finch specimens. It would be an incomplete account indeed of Darwin's revolution if it is attempted to be explained in terms of superior data collection or management. Living as we do in a time of a Big Data craze analogous somewhat to a time when a better naturalist was the one with a bigger specimen-collection and specimen-catalogue, many of today's generation of researchers in a pre-methodological phase in some disciplines but often with tonnes of data waiting to be 'mined', may have more in common than we admit with their counterparts from a time when 'formal theory' and 'empirical data' were yet to be duly applied to each other.

The large-scale adoption of Bayesian methods in formalizing the interplay as it were between theoretical assumptions and empirical measurement, coincided, fortuitously one might add, with other revolutions in measurement technology as was the case in the cognitive sciences. The undeniable progress in these disciplines around the 1980s is due to contributions from a larger milieu in which these operated of which the emergence of Bayesian methods was but a part. Speaking of the cultural milieu, something needs to be said also about the explosion in connectivity in this era, making possible collaborations (many of them interdisciplinary) that may have been impracticable earlier. It was after all a 'cultural difference' in how Darwin and Wallace operated that seems to have determined in large part their very different places in history.

Is there a case to treat Bayesian inference as more than an expedient machinery pressed into the service of examining theories in the light of evidence? Whether it is simply a machinery or a 'philosophy of science in its own right' is of less interest to a practicing scientist than its utility in coping with scarce and uncertain data and the throughput it affords in examining hypotheses with such data. On the 'machinery or philosophy' question that is really of more historical than scientific interest, I will hold off until reading what has been said here: Howson & Urbach, Scientific Reasoning: The Bayesian Approach (3rd edition, 2005)

That's enough words spent on a meta-review of the list itself and I guess it's about time I get to first a reading and then a review of some of the books in the list.GoodMorning
Reply


Possibly Related Threads...
Thread Author Replies Views Last Post
  Nihilistic Philosophy geetha 18 15,246 08-Apr-2014, 08:11 PM
Last Post: Kanad Kanhere
  A critique of Kuhn's philosophy of science Lije 0 4,545 30-Oct-2011, 01:21 PM
Last Post: Lije
  Why Should Freethinkers Study Philosophy? Debating a Christian Ajita Kamal 2 3,670 24-Jun-2011, 12:49 AM
Last Post: LMC
  Science & Philosophy - A dichotomy? vvjoshi 4 4,062 12-Jun-2011, 12:03 AM
Last Post: ARChakravarthy



Users browsing this thread: 1 Guest(s)