Header image from here.
Editor’s Note: Some of the ideas in this article, particularly in the beginning, were inspired by Huw Price’s essay “On the Origins of the Arrow of Time: Why There is Still a Puzzle about the Low Entropy Past.” It’s a great introduction to the subject discussed below, and it’s very readable for lay audiences who have little to no knowledge of physics or advanced math.
The arrow of time
Why does time flow in only one direction? The fact that time proceeds only from past to future is one of the few absolute certainties in life (or so it seems), yet it is also one of the greatest unsolved mysteries in physics.
The mystery deepened in the late 19th century when physicists discovered that the fundamental laws of the universe are time-symmetric. That is, they show no preference for a particular direction of time. For instance, imagine watching a video of two gas particles colliding with one another. It will usually be impossible to tell whether the video is being played forwards or backwards. The laws governing the motions of these particles, along with other simple systems like pendulums, will hold true even if, somehow, time is flowing in the reverse direction.
Yet many processes are clearly time-asymmetric. Gas will leave a balloon once its knot is untied, but it will never reenter and fill the balloon on its own; it is very easy to crack an egg, but virtually impossible to “uncrack” one. (1) Why is there so much time-asymmetry in the universe given the underlying time-symmetry in its laws?
An obvious answer in physics seems to come from the Second Law of Thermodynamics, which holds that the entropy of a closed system, such as the entire universe, will always increase. Entropy is generally defined for lay audiences as the measure of chaos or disorder in a system; however, as the preeminent thermodynamic physicist Ludwig Boltzmann realized in the late 1800s, it would be more accurate to describe entropy as a measure of the number of ways to rearrange the constituent elements of a system without changing its overall structure. For example, as Brian Cox explains in this (lovely) video, a sandcastle has a much lower entropy than a pile of sand.
Boltzmann’s reconceptualization of entropy is crucial for demonstrating why the arrow of time depends on the Second Law of Thermodynamics, since the law by itself does not explain how entropy could (time-asymmetrically) increase in a fundamentally time-symmetric universe. Statistically, there are far more ways for particles to arrange themselves into generic structures than to organize themselves into very unique ones; hence, it is very common to find piles of sand in nature, but it is exceptionally rare to encounter sandcastles that weren’t constructed by human beings. Thus, a system is much more likely to evolve into a high-entropy state than a low-entropy one.
However, the evolution of any (classical) physical system, whether into a high-entropy or a low-entropy state, is guided by the laws of Newtonian mechanics, which are, critically, time-symmetric. (2) As the physicist Johann Loschmidt pointed out to Boltzmann, it follows that for every system that starts at a low-entropy state in the past and develops into a high-entropy configuration in the future, there is an identical system that follows exactly the same trajectory, but from the future to the past. To be clear, physicists have been certain since the advent of thermodynamics that net entropy always goes up in the universe, but it was unknown why entropy rises towards the future and not towards the past.
There can only be time-asymmetry, then, if there is a boundary condition on the level of entropy in the universe at either the beginning or the end of its timeline, but not both. If the entropy in the universe was very low right after the Big Bang, then it must be the case that entropy increases from past to future. Alternatively, entropy would increase in the other temporal direction – and hence, time would flow in reverse – if overall entropy were high after the universe had sprung into existence. Since we empirically observe in our universe that entropy becomes higher in the future, physicists have sought to confirm that the universe started out with a low-entropy boundary condition.
And indeed, in the late 20th century, cosmologists discovered that matter was distributed very smoothly throughout the early universe. While this smoothness would seem to imply high entropy – the matter could be rearranged in many ways while still preserving the general structure of the universe – the dominant force in the universe at the time, gravity, was attractive, so the distribution of matter was extremely unstable. In fact, it was so unstable that, according to the physicist Roger Penrose, the chances that the early universe existed in this smooth arrangement are 1 in 10^10^123. These are astronomically low odds. It would have been much, much more likely for the matter in the universe to cluster together, but it didn’t. Why? I propose one answer in this blogpost, although it is worth noting that cosmic inflation is often regarded by physicists as another potential solution to the question.
The anthropic principle
If the universe had not been so smooth around the time of its inception, it would have been impossible for stars and galaxies to form. To clarify, the universe wasn’t perfectly smooth, otherwise it would have continued to stay that way for the entirety of its history. Rather, there were a few irregularities dispersed at exactly the right spatial intervals to permit the development of any large-scale structure of matter. It seems like the universe began at precisely the right level of entropy to give rise to the kinds of cosmological systems that could eventually accommodate the evolution of intelligent life. In other words, the universe was either a monumentally improbable fluke or it was finely-tuned for biological creatures like us and the other living entities in the world. (There is another possibility, which I will discuss later: we actually live in a massive ‘multiverse,’ in which the overwhelming majority of the other universes do not have the necessary conditions for life.) This “fine-tuning” argument is still quite controversial among physicists, but I find it significantly more compelling than the notion that low entropy in the early universe was nothing more than a statistical anomaly. As the physicist Lee Smolin once wrote, “In my opinion, a probability this tiny [as the probability of a low entropy boundary condition on the universe] is not something that we can let go unexplained. Luck will certainly not do here; we need some rational explanation of how something this unlikely turned out to be the case.”
The concept of fine-tuning is similar to the anthropic principle, which comes in two flavors. The weak anthropic principle (WAP) states that the universe currently exists in such a way that it has developed hospitable conditions for life. WAP is obviously true insofar as we are alive. The strong anthropic principle (SAP), on the other hand, makes a much more forceful and profound claim: the universe must have evolved in such a way that it produced intelligent life.
It would be very misguided to seek “scientific proof” of SAP, and even if it weren’t, the most scientifically compelling evidence for SAP, i.e. the low probabilities of the initial conditions of the universe, has arguably already been discovered by cosmologists. It doesn’t make sense to search for empirical insight into the “intentions” of the universe; through their research, physicists can only gain an understanding of how the universe executes these intentions, if it has any at all, or how the universe is blindly and arbitrarily following certain laws that were randomly established. Why the universe has evolved to its current state – and, more broadly, why anything happens at all – will always remain, at least on some level, a mystery to scientists. Later in this blogpost, I try to offer a philosophical defense of SAP.
Physicists may object to SAP (and, implicitly, my analysis in the prior paragraph) by arguing that the way the universe has historically evolved indicates that it lacks the intentions that the principle is seeking to impute to it. In particular, if the universe’s goal were to create life, these critics might ask, then why did it go to the trouble of generating so many galaxies that are absolutely bereft of conscious existence? Why is the universe so inefficient at producing conscious beings? To quote Huw Price, a philosopher of science, “Life as we know it doesn’t seem to require an early universe which is smooth everywhere, but only one which is smooth in a sufficiently large area to allow a galaxy or two to form (and to remain relatively undisturbed while intelligent life evolves).” It is possible to resolve this objection in a manner that is entirely consistent with the notion that the universe has intentionality; perhaps the universe has plans to, over many billions of years, develop those other regions of the universe into areas that contain intelligent life. But this is mere speculation; even the most advanced models in astrophysics cannot yet predict where life will evolve and when. However, even if the other galaxies in the universe remain fallow for the rest of eternity, it is nevertheless the case that the universe must have begun in a state of low entropy if life were to eventually form. Regardless of whether the universe took the most energy-efficient route to intelligent life once those starting conditions had been determined, the fact that those conditions were even established in the first place remains an unexplained miracle unless we accept the Anthropic Principle (or some variant of it) as truth.
Biological life depends on the low-entropy state of the early universe for another reason besides its necessity for the formation of stars and galaxies. The time-asymmetry that emerged from it is a precondition for our capacity to act as agents operating in the world. An essential feature of any living system is its ability to receive a set of inputs in order to produce outputs that are conducive to its survival. Even an organism as simple as a plant takes sunlight as an input in order to make glucose, which serves as its source of energy. The “in-order-to” structure of this process inherently implies a direction of causality; the inputs manipulated by a biological agent act as causes, and these must always precede their effects, which are the resulting outputs. As Price writes,
“The origins of causal asymmetry thus lie in our experience of doing one thing to achieve another—in the fact that in the circumstances in which this is possible, we cannot reverse the order of things, bringing about the second state of affairs in order to achieve the first. This gives us the causal arrow, the distinction between cause and effect. The alignment of this arrow with the temporal arrow then follows from the fact that it is normally impossible to achieve an earlier end by bringing about a later means.”
Furthermore, if the universe were time-symmetric (that is, if there were no boundary conditions on its levels of entropy), then causality would be impossible. Photosynthesis cannot cause the production of glucose if it is true that the production of glucose is causing photosynthesis; in general, event A cannot be said to cause event B if B also causes A in the reversed time direction. Obviously, there are feedback processes in which a certain event can act as both a cause and an effect. But even in these cases, the cause becomes an effect at a later moment in time. At any given moment, a particular event can only be either a cause or an effect. If it were possible to reverse time, however, the event could be both simultaneously, thereby rendering any notion of causality incoherent.
Therefore, causality can only be a real phenomenon if time is asymmetric, and furthermore, life as we know it could not exist unless it is possible to cause change in the environment. However, this doesn’t really amount to a strong argument in favor of SAP, since the universe didn’t need to evolve with the intention of producing life in order to establish causality. Plenty of processes are causal even if living beings are not there to experience them. For instance, consider an earthquake that occurred on Earth billions of years ago, before single-celled microorganisms had even emerged. Despite the lack of intelligent life on the planet, it nevertheless appears to be true that the shifting of seismic plates caused the earthquake to happen.
Yet one could argue that there are only correlations, rather than causal events, in a universe without living, conscious entities. The philosopher David Hume famously contended that the phenomenon we refer to as causation is really just a “constant conjunction”; if we perceive that two events always occur together, then we will deem them to be causally related. We notice that movements in tectonic plates will consistently be followed by earthquakes, hence we know that the former event is the cause of the latter. Causation, therefore, is nothing more than a mental abstraction that intelligent agents like us project onto a succession of events. Outside of the mind, there are only events that precede and succeed one another.
This argument is entangled with many deep and well-worn philosophical questions about the ontological status of mental concepts; in particular, are they just as real as the phenomena that we observe in the outer world? Furthermore, insofar as “events” are also abstractions, is it even fair to say that events, as far as we can conceptualize them, are somehow more “objective” than the causal relations between them? These are tough challenges, and I will not attempt to address them here since they would demand a much longer exposition of the vast philosophical literature about them.
Nevertheless, there are valid reasons for believing that causation is a phenomenon that only intelligent biological agents can access or experience, rather than an objective feature of the physical universe. The idea that a cause begets an effect, and not the other way around, is derived from our subjective feeling that causes are already determined, whereas we can freely choose the effects that we subsequently bring about. (3) According to Price, “The difference between the fixity of the inputs from the past and the openness of the outputs to the future is a feature of the experience from the inside—a feature of what it feels like to be an agent.” Subjectivity – and, more broadly, consciousness – play a crucial role in this perspective. Lava flowing on the Earth’s surface cannot be aware that it is causing the sediment around it to melt. Unless witnessed from a subjective viewpoint, the melting of the rock merely takes place after the lava starts moving. The notion that the former depends on the latter can only arise from the consciousness of an intelligent, biological agent that can feel the qualitative difference between a certain past and an uncertain future.
Why does consciousness need to be involved? Couldn’t a non-conscious entity that can nevertheless perform computations about time, such as an artificial intelligence (AI), be capable of distinguishing between cause and effect? No, AI, including machine-learning systems that seek to identify unknown data on the basis of known data received in the past, cannot tell time in the way that biological creatures do. These systems may receive large datasets as inputs to output useful predictions, but they don’t have any real sense that the inputs are fixed and the outputs are somehow “subject to their control.” They are just manipulating data in accordance with the algorithms that have been programmed into it, with no conception of the past as something that they have left behind and the future as a time that they have yet to experience. One might note that simple biological organisms like plants also lack this subjective awareness of time, which is why I’m actually more inclined to believe that everything experiences the passage of time. However, in this view, there would be certain entities that have a greater or more sophisticated consciousness of time. The need to survive, which sets apart all living beings (including plants) from inanimate objects, engenders goal-directed behavior, such that inputs received from the past are manipulated with the purpose of improving future well-being. Only entities that have some fundamental desire for self-preservation can have a meaningful conception of “purpose,” so only they can phenomenally understand what it means for the future to causally depend on the past. (Although many people would be skeptical of ascribing sentience to them, I would argue that plants do have this desire as well, but at such a rudimentary level that we cannot really empathize with them.)
Even if we accept that causality presupposes a consciousness that can conceive of it, we are still left with a fundamental problem: how could causality be so significant that the universe went through all the effort of creating conscious, living beings in order to realize it as a meaningful phenomenon? What value does causality have to the universe? First, it is worth noting that the answer to this question must be philosophical in nature, since empirical findings in physics cannot be used to adjudicate concerns about what the universe values, if anything. Questions about these values are tied up with the subject of why the universe behaves in the way that it does, whereas physics can only predict and explain how the universe behaves.
I still have not yet completely formulated my stance on the “cosmic importance” of causality, but I suspect that it is closely related to another metaphysical idea that I am interested in developing: self-referential cosmopsychism. Roughly, this view holds that the universe is a single consciousness that is becoming aware of itself through the experiences of the subjective observers inside it. (4) Self-referential cosmopsychism is beautifully encapsulated in this quote by the spiritual teacher Alan Watts:
“Through our eyes, the universe is perceiving itself. Through our ears, the universe is listening to its harmonies. We are the witnesses through which the universe becomes conscious of its glory, of its magnificence.”
In essence, the universe must establish causality in order to imbue its own existence with purpose. Unless it is possible to even cause an outcome in the first place, it is meaningless to say that the Big Bang “brought about” the existence of the universe, or that the smooth distribution of matter in the early cosmos “resulted” in the arrow of time. In a universe without causality, there would merely be events that happen in a particular order. There would be no reason that events happen because the occurrence of any phenomenon could not be said to depend on a cause, even a purely physical one. Since we understand the concepts of cause and effect, we know that we are alive because of the universe. Furthermore, we can attribute an intention to the universe only because it makes sense to us that the universe would be capable of causing an event (such as the emergence of life) in order to execute some outcome. In other words, any intention that the universe may have presupposes that there is some meaningful conception of intentionality, which could only belong to entities that know what it means to cause a desired effect. Unless we posit that there are supernatural beings that can apprehend causality, the evolution of life would be a prerequisite to the fulfillment of whatever ultimate intention the universe has. Given the incredibly slim chances of our existence – recall the 1 in 10^10^123 odds of establishing a low-entropy boundary condition on the universe – it seems highly plausible to claim that the universe does have an intention and that we play a crucial role in realizing it. In fact, in accordance with the framework of self-referential cosmopsychism, the universe is, through us, becoming aware of its own intention over time; it does not ‘know’ this intention in advance.
I will conclude by noting that, according to my perspective, the evolution of the universe has the character of a paradoxical loop. (5) If the concepts of causality and intentionality depend on the existence of conscious beings yet there was no living consciousness at the start of the cosmos, then how could the universe have begun with the intention of producing life? Could the universe have evolved with the intention of creating entities that make intentionality possible? Such a view would likely seem incoherent, since we cannot reach back into the past to determine the universe’s original intention. After all, one of the aims of this blogpost has been to explain why we only perceive the flow of time in one direction.
But if the universe must have produced life in order to fulfill its ultimate intention, then how could it have evolved any other way?
tl;dr: The universe began in a state of low entropy, hence time only flows in one direction. The experience of the “asymmetry” of time is a precondition for the ability of conscious beings to understand causality. The universe was likely fine-tuned to have a low-entropy boundary condition so that it could later evolve to produce conscious beings who essentially “create” the concept of causality. Furthermore, this realization of causality permits the universe to become aware of itself over time.
(1) Side note: Chemists at UCI recently discovered a way to “unboil” an egg.
(2) I don’t know enough about quantum mechanics to know whether this temporal symmetry is true for quantum processes as well.
(3) Note that I’m implicitly equating biological intelligence with consciousness in this essay. While I do believe that all living beings are conscious, I do not think that intelligence and consciousness are one and the same, nor do I think that one must arise from the other. One does not have to subjectively experience a phenomenon in order to perform cognitive computations about it. My justifications for this claim should be saved for a different blogpost, since they are outside the scope of this particular article.
(4) It would be more accurate to say that the universe is a non-dual (rather than a single) consciousness, although the distinction between “non-dual” and “single” awarenesses will also have to wait for a different essay.
(5) I have discussed these “strange loops” in previous blogposts.