### Looking for references: phenomenology of probability

#### by Sebastian Benthall

A number of lines of inquiry have all been pointing in the same direction for me. I now have a question and I’m on the lookout for scholarly references on it. I haven’t been able to find anything useful through my ordinary means.

I’m looking for a phenomenology of probability.

Hopefully the following paragraphs will make it clearer what I mean.

By *phenomenology*, I mean a systematic account (*-ology*) of lived experience (*phenomen-*). I’m looking for references especially in the “cone” of influences on Merleau-Ponty, and the “cone” of those influenced by Merleau-Ponty.

By probability, I mean the whole gestalt of uncertainty, expectation, and realization that is normally covered by the mathematical subject. The simplest example is the experience of tossing a coin. But there are countless others; this is a ubiquitous mode of phenomenon.

There is at least some indication that this phenomenon is difficult to provide a systematic account for. Probabilistic reasoning is not a very common skill. Perhaps the best account of this that I can think of is in Philip Tetlock’s *Superforecasting*, in which he reports that a large proportion of people are able to intuit only two kinds of uncertainty (“probably will happen” or “probably won’t happen”), another portion can reason in three (“probably will”, “probably won’t”, and “I don’t know”). For some people, asking for graded expectations (“I think there’s a 30% chance it will happen”) is more or less meaningless.

Nevertheless, all the major quantitative institutions–finance, telecom, digital services, insurance, the hard sciences, etc.–thrive on probabilistic calculations. Perhaps there’s a concentration here.

The other consideration leading towards the question of *phenomenology* of probability is the question of the interpretation of mathematical probability theory. As is well known, the same mathematics can be interpreted in multiple ways. There is an ‘objective’, frequentist interpretation, according to which probability is the frequency of events in the world. But with the rise of machine learning ‘subjectivist’ or Bayesian interpretations became much more popular. Bayesian probability is a calculus of rational subjective expectations, and transformation of those expectations, according to new evidence.

So far in my studies and research, I’ve never encountered a synthesis of Merleau-Pontean phenomenology with the subjectivist intepretation of probability. This is somewhat troubling.

Is there a treatment of this anywhere?

Its a fascinating question that Ive also come across. I have no idea of the intellectual lineage wrt to Merleau-Ponty but the area of math / stats education and the work of folks like Paul Cobb, Richard Lehrer, Cliff Konold and others might be relevant. One project I remember used computational simulation to help learners develop an intuitive sense of statistics. Heres a hidden gem I found several years ago that I think you will especially like: https://dspace.library.uu.nl/bitstream/handle/1874/893/full.pdf

Oh, awesome! Thanks for these references. They are absolutely relevant and helpful.

Hi, that’s funny, I was trying to find the exact same thing and I landed here by googling “phenomenology of probability”. I’m a postdoc working on the subjective Bayesian interpretation of quantum mechanics (aka QBism). Lately we’re really excited about phenomenology, especially Merleau-Ponty’s stuff, because we think it can help us resolve some of our deep metaphysical problems.

Of course, being quantum physicists, the first thing we want to know is what the phenomenologists have to say about probability. And … we’ve got nothing! Weirdly it seems like they haven’t thought about it much. Or maybe they only wrote about it in French or German. I’m doing some digging now for a paper I’m writing, so if I do find anything, I’ll post it here.

Why did you become interested in this? I know what my excuse is, but I can’t imagine why someone else would think to go down this rabbit hole.

I’ll admit that I’ve taken a very long time to reply because I’m not sure I can explain my answer easily.

Let me try.

I have a background in Bayesian cognitive science, and continue to be interested in the philosophical foundations of my work on cybersecurity and the data economy. For example, this paper:

I think Fred Dretske’s work connecting epistemology and information theory is very promising. But as he is naturalizing epistemology, he is not engaging the “first person” aspects — phenomenology.

In my field, there’s been a bit of tension between the formal methods people and those that use anthropological or humanistic methods, which draw ultimately from a more “first person”, perhap phenomenological, base.

I think there’s good work that connects the two domains at a high level in Eduardo Kohn’s anthropology, via Peircean semiotics and Terence Deacon’s biosemiotics. But nothing has nailed this down into what I see as the two intellectually dominant disciplines: STEM, specifically computer science’s merger with Bayesian statistics; and the continental tradition, which is dominant in European law and which via the Brussels Effect has been shaping global data policy.

So the phenomenology of probability would get to the epistemic root of some key research problems in my area. This is of course an extremely small minority position.

Does that have something to do with Bayesian interpretations of quantum physics? Yes, I expect so.