The argument of this paper is that an arrow of time associated with information processing systems cannot be deduced from thermodynamic arguments. The thermodynamic arrow is insufficient to entail the computational arrow. Any sequence of logical operations in an entropy increasing universe is physical possible in an entropy decreasing universe. Landauer's Principle, as it is commonly stated, assumes statistical mechanical principles that are equivalent to being in an entropy increasing universe. If one changes those assumptions, so that one is in an entropy decreasing universe, a critical inequality in Landauer's Principle in reversed. The physical implementation of logical operations, which increase entropy, do so, not by virtue of any inherent properties of the logical operation, but by virtue of being in an entropy increasing universe. If the same logical operation is performed in an entropy decreasing universe, it is entropy decreasing. As a result, entropy decreasing universes are not inherently hostile to the acquisition, persistence or utilisation of information.

In principle, the operation of acquiring information can be made thermodynamically reversible. This is precisely one of the main insights of Landauer's work on the thermodynamics of computation: a measurement can take place without generating heat.

Landauers principle, while perhaps obvious in retrospect, makes it clear that information processing and acquisition have no intrinsic, irreducible thermodynamic cost.

If the acquisition of information can take place in a thermodynamically neutral manner, it can take place in an entropy decreasing as easily as an entropy increasing universe.

While any information gathering and utilising system will ultimately cease to function in an entropy decreasing universe that reaches final extremal entropy state, this doesn't seem sufficient to rule out such systems. Firstly, the decrease in entropy is due to the decorrelation that comes about from losing microcorrelations. It is of a different kind to the macrocorrelations that arise during the acquisition of information. Secondly, on the timescales during which information gathering and utilising systems work, between the low or high entropy extremal starting and ending points, there seems nothing to directly prefer IGUS over RIGUS. Thirdly, if the effect of the initial or future boundary conditions is screened off by the local entropy gradient, the no correlation, no interaction argument does not seem to be
applicable, as such an entropy gradient can exist in a situation with no initial orfinal boundary condition.

The suggestion is made that an entropic arrow of time will never be found in processes that can be defined solely in terms of a succession of Quasi-Static Equilibrium States. Information processing can be so defined. If the psychological arrow of time is to be aligned with the thermodynamic arrow, it cannot be through the information processing properties of the brain. It may be through the biochemical structures that arise in Non-Equilibrium Steady State processes, but if so, it is certainly not through any information processing characterisation of such structures. This would seem to imply that at least one aspect of conscious experience cannot be logically supervenient on the states of a computer. If instead the psychological arrow of time does indeed arise out of information processing properties, this would mean that the psychological arrow is logically independant of the thermodynamic arrow of time.

Thermodynamic entropy can be a tricky area even for physicists. Suppose you fell through a wormhole into a part of the universe where everything went backwards. Broken eggs would put themselves together again, if you had a beer you would feel drunk beforehand and thirsty afterwards. But Stephen Hawking theorised you wouldn't be able to tell this had happened, because you'd start seeing and remembering things backwards too - your past would become your future. Now Dr Owen Maroney, a postdoctoral research fellow working in the areas of Physics, Philosophy and Time at the University of Sydney, says this is not correct, we can remember in a reversed universe and it's all to do with the heat coming from your computer.

Already working on the problem of heat generated by computers, Maroney went to a lecture at Imperial College, London, on computers, the direction of time, and their relation to our memories and perceptions, when he started wondering what would happen if you calculated the heat given off by a computer in the reversed universe. "People had assumed that it wouldn't change things" he said, "but I realised that's all it was - an assumption."

If you break an egg, entropy goes up. You can't put the egg back together, because physics says entropy can't decrease. The theory is that in the reversed universe, entropy has to go down, so broken eggs can get put back together again. "Computers get hot. That increases entropy as well. So Hawking had said that this means computers can't work in the reversed universe. They'd have to start running backwards. The brain is a kind of computer, so you'd go into reverse too and still end up seeing the egg breaking apart."

But until now no-one had ever done the calculation. Maroney worked it out and found that it works in the opposite way to that expected. "In the reversed universe, computers will absorb heat instead. They'll get cold. That makes entropy go down, so they can work after all. And that means so can you," explains Maroney.

So Humpty Dumpty might have a chance after all. But there may be a problem: all this has only been worked out for classical computers. With a quantum computer things may be different, and Maroney is still working on the details. "It is interesting given my research could have an impact on quantum computing that it's mainly quantum physicists who argue with my findings," he says with a bemused look. Maroney's paper "Does a Computer Have an Arrow of Time?" is to be published in the February 2010 issue of Foundations of Physics. Maroney is a Post-Doctoral Research Fellow with the Perimeter Institute Australia Foundations Collaboration.