Abstract
It is commonly thought that the statistical mechanical reversibility objection implies that our putative records of the past are more likely to have arisen as spontaneous fluctuations from equilibrium states than through causal processes that correctly indicate past states of affairs. Hence, so the story goes, without some further assumption that solves the reversibility objection, such as the past hypothesis, all our beliefs about the past would almost surely be false. This claim is disputed and it is argued that at least some of our records of the past can and should be thought to be veridical because the intentional contents of records are not included as part of their statistical mechanical description. The fact that the present state of the world around us coheres so well with the way we would expect it to be if our records were veridical provides good evidence for the claim that they are produced via a common causal structure.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
We can think of a macrocondition as being a list of the macroscopic properties of a system that we are capable of observing such as temperature, pressure, volume, etc.
In the case of Newtonian mechanics, this amounts to reversing the velocities of each particle in the system.
Given that this is how Albert poses the problem, this is the framework I adopt. There are several ways of posing the argument more generally. One might consider a Boltzmann’s brain scenario, where instead of the entire present macroscopic state of the universe, one considers it most likely that only one’s own brain spontaneously fluctuated into existence. Similarly, one might question why the laws of motion should be taken as a given, as our confidence in these laws is presumably underwritten by the veracity of our records.
At least in the case of predicted physical phenomena.
This is essentially Albert’s (2000) description of what a record is.
One might dispute this understanding of the content of records being externally individuated, instead claiming that records are individuated by their causal histories. However, since this paper addresses an inferential problem, this conception should be unproblematic.
This does not preclude a completely thermalised state from having some representational content. There is no a priori requirement that records need to be low entropy states, as Earman (1974) forcefully argues.
In the correlated case, the information-theoretic entropy is given by H(X) + H(Y|X) = H(X), while when uncorrelated, the entropy is expressed by H(X) + H(X|Y) = H(X) + H(Y).
This may also measure the physical complexity associated with the string of memory cells (Adami and Cerf 2000). Of course, the fact that the physical systems that comprise our own memories or a computer memory have any representational content at all depends on various complex organisational features of our brains and the computer that are not themselves accounted for by merely assigning to these systems a thermodynamic entropy. Adami and Cerf treat this feature, not incorporated into the argument here, through the notion of a “stable site” that encodes the instructions for recovering the content of the string.
Again, this presumption is made for the sake of simplicity. Clearly, different record bearing states may, in principle, have differing entropic values though this does not change the nature of the above analysis.
For simplicity and ease of presentation, the following analysis is presented at the level of personal and individualised records. However, it should be recalled that the sceptical problem dealt with here is not intended to be posed as a solipsistic one, but one that includes all record-using agents in the epistemically accessible macrostate of the universe.
Nothing is intended to hinge on the choice of the tipping point at 0.5, but is merely chosen for convenience.
If we are in possession of a probability distribution over the space of possible memory contents and these probabilities are independent of the statistical mechanical probability distribution, then P(M) is factorisable: P(M) = PR(M)PSM(M). This may not generally be true, but as long as such correlations are weak, the argument should go through.
We assume that any non-maximal entropic state has negligible measure on the usual phase space representation.
U is now the past macrostate 5 min ago of the table and H its present macrostate, which are nearly identical.
This illustrates the point that the veracity of our records depends on contingent facts about the present macrostate of the world. If the universe’s macrostate were without these apparent correlations, the reversibility argument would go through.
This cannot generally be true. Clearly some records have representational content in part due to their thermodynamic description, and there might be some correlation between the thermodynamic states of the record and the recorded system. The present point is that there is no necessary connection, not that there never is or cannot be one.
This is necessary since the physical constitution of different memories contents might be differently realised, and thus not all equally probable.
References
Adami, C., & Cerf, N. J. (2000). Physical complexity of symbolic sequences. Physica D: Nonlinear Phenomena, 137, 62–69.
Albert, D. (2000). Time and chance. Cambridge, MA: Harvard University Press.
Bennett, C. (2003). Notes on Landauer’s principle, reversible computation, and Maxwell’s Demon. Studies in History and Philosophy of Modern Physics, 34, 501–510.
Earman, J. (1974). An attempt to add a little direction to “The Problem of the Direction of Time”. Philosophy of Science, 41, 15–47.
Earman, J. (2006). The “Past Hypothesis”: Not even false. Studies in History and Philosophy of Modern Physics, 37, 399–430.
Hemmo, M., & Shenker, O. (2007). Prediction and retrodiction in Boltzmann’s approach to classical statistical mechanics. http://philsci-archive.pitt.edu/archive/00003142/.
Horwich, P. (1987). Asymmetries in time. Cambridge: MIT Press.
Merleau-Ponty, M. (1945). Phénoménologie de la Perception. Paris: Gallimard.
Parker, D. (2005). Irreversible thermodynamics: Does the big bang explain what it purports to explain? Philosophy of Science, 72, 751–763.
Price, H. (1996). Time’s arrow and archimedes point. Oxford: Oxford University Press.
Reichenbach, H. (1956). The direction of time. Los Angeles: University of California Press.
Savitt, S. (1990). Epistemological time asymmetry. In PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 1:317–324.
Shenker, O. (2000). Logic and entropy. http://philsci-archive.pitt.edu/archive/00000115/.
Sklar, L. (1993). Physics and chance. Cambridge: Cambridge University Press.
Smart, J. C. C. (1967). Time. In P. Edwards (Ed.), Encyclopedia of philosophy. New York: Macmillan.
Winsberg, E. (2004). Can conditioning on the “Past Hypothesis” militate against the reversibility objections? Philosophy of Science, 71, 489–504.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Parker, D. Was There an Ice Cube There or Am I Just Remembering It?: Does the Reversibility Argument Really Imply Scepticism About Records?. Erkenn 80 (Suppl 3), 587–603 (2015). https://doi.org/10.1007/s10670-015-9768-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-015-9768-4