DEEP in the Amazon rainforest, a tree frog sits on a log watching a fly. A genetic fluke has furnished the frog with a brain that perceives its surroundings as they were a second ago. When the fly comes within range, the frog lunges. But, with its out-of-date observation, it misses. Weakened by a rarely sated hunger, the frog falls off the log and dies.
It鈥檚 a sad story. But if you think it is completely fanciful, think again. There is nothing in the laws of physics that says all creatures have to process data about their environment in the same way as we do. A 鈥渂ehind the times鈥 perception like that of our deceased frog is only ruled out by the handicap it imposes. 鈥淣atural selection has equipped people and frogs to experience the world in the most effective way for their survival,鈥 says James Hartle of the University of California, Santa Barbara. 鈥淎 frog that calculates the trajectory of a fly from the most recent data, eats; one that doesn鈥檛, starves.鈥
Perhaps this seems mere common sense. In a way it does to Hartle too. So why is this heavyweight cosmologist 鈥 his collaborators include Stephen Hawking, Murray Gell-Mann and Steven Weinberg 鈥 worrying about frogs?
Advertisement
Because he believes it goes to the very heart of why we perceive time as we do. The constraints that evolution imposes on our perception of time could help us understand time鈥檚 true nature. 鈥淥ur powerful sense that there is a 鈥榥ow鈥 and that time 鈥榝lows鈥 from the past, through the present, to the future, has survival value,鈥 he says. 鈥淚t is the only plausible explanation, since none of these concepts actually appear in Einstein鈥檚 special theory of relativity, our most fundamental physical description of space and time.鈥
After many years of worrying about time, Hartle has put aside a little of it to consider what evolution might tell us about past, present and future. After adding in the concerns of a few alien civilisations and constructing a robot that has premonitions, he believes things are starting to become clearer. Really, he does.
Even a little thought reveals that the idea of a 鈥渇low of time鈥 is a nonsense, Hartle says. 鈥淪omething which flows, changes with time,鈥 he points out. 鈥淏ut how can time change with time? It鈥檚 a logical impossibility.鈥 This is made explicit in special relativity. According to Einstein, space-time is essentially a four-dimensional landscape (with three dimensions of space and one of time) in which all the events in the history of the universe 鈥 from the big bang into the far future 鈥 are laid out exactly as if they are pre-ordained. Nothing flows.
But, even if the flow of time does not appear in our description of reality, surely it is still possible to agree on what we all mean by past, present and future? 鈥淪adly, no,鈥 says Hartle.
It is a well-documented, if rather strange, facet of relativity: time is not an independent phenomenon, but inextricably linked with space. And if people within this 鈥渟pace-time鈥 are moving through space relative to one another, the distinction between space and time becomes blurred: they cannot agree on what 鈥渘ow鈥 means. 鈥淚n Einstein鈥檚 universe, there can be no such thing as a universally agreed past, present and future,鈥 Hartle says. 鈥淲ithin the context of space-time, the very concepts are meaningless.鈥
So, how to make sense of it? The answer, he says, lies in the concept of information: it is information 鈥 and the way we process it 鈥 that gives us a sense of past and future.
Hartle reached this conclusion by considering what he terms a subsystem of the universe, defined as something that gathers and uses information. Human beings, both individually and collectively (as members of civilisations), fit into this definition, as do frogs, and even robots. It was the physicist Gell-Mann who gave these subsystems a name 鈥 he called them 鈥渋nformation gathering and utilising systems鈥 (IGUSs).
The mechanism behind the IGUS has turned out to be a surprisingly useful concept (see Diagram). First, information from its environment is held in an input register. But this register has a limited capacity, and as more information comes in from the environment, the information is passed to a memory register to clear space for new input. The IGUS might have many memory registers, along which it can shuffle the past information. Eventually though, it will have to be dumped.
Of course, to dump this past information without using it in any way would be a waste, so before that happens it is passed to other parts of the information processing system. In addition to the registers, the IGUS has a 鈥渟chema鈥, a simplified model of its environment with rules culled from its past experience, which tells it how to behave in particular circumstances. It also has a computer that works out how it should react to its surroundings, based on the rules stored in the schema. The computer carries out two distinct kinds of computation: conscious, which makes decisions; and unconscious, which updates the schema.
There are many ways in which information can pass between all these components, Hartle points out, and they create very different behaviours in the IGUS. The most familiar idea 鈥 what we would consider normal 鈥 is to allow conscious to focus on the present (the input register), and unconscious on the past. 鈥淭his distinction is important because it is the suggested reason we consciously experience the present but remember the past,鈥 Hartle says. He contends that this simple set-up mimics some of the key features of human perception. For a start, because an IGUS focuses its attention on the most recently acquired image, the present is a specific thing with a special immediacy.
It is the passage of images between registers, until they are erased and 鈥渇orgotten鈥 that creates the impression of a flow of time. 鈥淪omething analogous to the flow of information from register to register happens in our brains, and this is what ultimately gives us our sense of time passing,鈥 Hartle says. The perception of time as flowing is perfectly compatible with Einstein鈥檚 space-time as long as the 鈥渇low of time鈥 is in fact a 鈥渇low of information鈥. Furthermore, he says, for an IGUS the past and future are qualitatively different, in that the past in the registers is 鈥渞emembered鈥 while the future is 鈥減redicted鈥 as the output of computation.
Equipped with this set-up, Hartle imagines what happens when an IGUS observes an object in its environment 鈥 when an image of a cheeseburger appears in the input register, for example. 鈥淭he computer consults the schema, which has abstracted rules from a previous experience 鈥 previous visits to burger restaurants, for example 鈥 and realises 鈥楬ey, I like cheeseburgers鈥,鈥 Hartle says. 鈥淭he IGUS therefore decides to buy a cheeseburger. Or perhaps the schema contains information on the fat content of burgers, which overrides the liking of burgers, so the computer decides not to buy a cheeseburger.鈥
This seems reasonable and familiar, showing that we use past information to inform decisions to be made in the present that might change the future. But what about the problem of the common now; what about the fact that, in space-time, it is not always possible for observers moving relative to each other to agree on the past-future ordering of events? That doesn鈥檛 fit with our experience; on Earth we IGUSs define 鈥渘ow鈥 quite easily.
Well, says Hartle, imagine two or more of the IGUSs embedded in 4D space-time. In this situation, although there can be no universally agreed 鈥渘ow鈥 in space-time, there can nevertheless be an approximate, 鈥渓ocal now鈥, given three restrictive conditions. The first condition is that the observers are separated by a short distance compared with the distance light can travel in the time of the observed events. The second condition is that the observers must be moving relative to each other at significantly less than the speed of light. And the last condition is that the time for perception 鈥 that is, for processing the new information in the input register 鈥 must be short compared with the time over which interesting features of the environment change. 鈥淭hat is probably necessary for a creature to function,鈥 Hartle says. 鈥淎 frog would have a hard time if it took longer to perceive a position than for the fly to move.鈥
All three conditions are easily satisfied for human beings on Earth, but Hartle points out that the first condition would not be satisfied for an IGUS that comprised an alien civilisation spread across a whole galaxy. 鈥淭here would be no point in defining a 鈥榥ow鈥 on a planet at the centre of the galaxy when light would take 60,000 years to take knowledge of it to a planet on the periphery,鈥 he says. 鈥淐learly, such a civilisation would need to organise its time differently from us.鈥
And that鈥檚 when the robots come in. Hartle imagined building robot IGUSs according to his own peculiar blueprints; because they are so flexible in their set-up, they can help determine whether there are other ways in which creatures could organise their experience of time 鈥 ways that are still consistent with the basic laws of physics. 鈥淚t is possible to imagine an IGUS that organises its experience much like we do,鈥 he says. 鈥淏ut, importantly, it is also possible to imagine IGUSs which organise their experience in quite different ways.鈥
To explore this, Hartle has come up with three variations on his basic IGUS. The 鈥渟plit-screen鈥 robot focuses on not one but two moments of time, 10 seconds apart, so it has two 鈥渘ows鈥. The 鈥渁lways-behind鈥 robot, like our poor deceased frog, sees the world as it was a few seconds ago. And the 鈥渘o-schema鈥 robot must calculate its next move from the contents of all its registers, because they all feed directly into the conscious computation.
鈥淲ould any of these be viable?鈥 Hartle asks. Well, perhaps, but they all seem to have problems 鈥 to us, at least. The split-screen robot would be wasting valuable conscious focus on inessential information in the past. As we have seen, the always-behind robot would starve to death. And the no-schema robot would waste its computational resources.
Hartle concludes that such variant ways of organising experience, should they ever arise, would be promptly weeded out by natural selection. 鈥淥ur time sense is determined not by physics alone but also by biology,鈥 he says. 鈥淚f this is right, any extraterrestrials we meet will experience the world the same as us, sharing concepts of past, present and future, and the idea of a flow of time.鈥
He points out, however, that if the laws of physics were different they would have given rise to creatures that organised their experience differently from us. 鈥淪ay the force on a body depended on its position now and 10 seconds ago,鈥 he says. 鈥淣atural selection would favour the evolution of a split-time creature.鈥
But, though natural selection based on the laws of physics has made humans process information in the way we do, there must be other possibilities beyond what he has imagined so far. After all, we are free to build robots any way we like.
Indeed, the satellites that make up the Global Positioning System could be considered as an IGUS that is programmed to perceive time in a peculiar way. People using the GPS rely on receiving signals from clocks on the satellites. By comparing the different amounts of time the various satellite signals take to reach it, the user鈥檚 receiver then determines its position on Earth. The system depends on the satellite clocks being synchronised but, since they are all accelerating and moving at high speeds relative to the receiver, relativity dictates that they all have different perceptions of 鈥渘ow鈥. For the system to operate, the GPS engineers had to design a particular meaning of 鈥渘ow鈥, adjusting each of the satellite clocks鈥 ticking rates, and programming receivers to perform the necessary relativistic corrections. In other words, they tinker with the system鈥檚 perception of time.
But engineers could push things further, Hartle suggests. It might even be possible to build a robot which organises its experience back-to-front, one that remembers the future and predicts the past. 鈥淭hough it is difficult, it is not impossible in principle,鈥 he says.
To create a robot with a reversed perception of time, Hartle simply has to reconstruct information records about the robot鈥檚 environment that have been destroyed, and then feed them through the robot鈥檚 registers backwards in a way that would make it look 鈥 to the robot at least 鈥 as though information was coming from the future. Because it travels through the memory registers, it would seem like a memory of the future.
Imagine, for example, a time-reversed robot observing an egg falling to the ground. In the usual time order, the record of the destruction of the egg is erased some time after the egg hits the floor. How would a reversed robot and egg look to us? We would see the record unerase and then, later, the egg fall. 鈥淭hus, in between, the robot has records of the future,鈥 Hartle says.
In practice, Hartle points out, achieving this would be much more tricky even than sticking the contents and shell of a broken egg back together. The robot would have to operate in an environment showing this reversed entropy: that would mean reversing the velocity of every particle of matter and photon of energy in the robot鈥檚 neighbourhood. 鈥淪ince we have a system of matter interacting with light, it would be necessary to deal with every molecule and photon within a radius of 60,000 light years,鈥 Hartle says. 鈥淲e lack the technology to do that.鈥
And so, for the moment, he is contenting himself with the conclusion that every creature that we know about is likely to organise its time in the same way as we do. But maybe, out there in the far reaches of space, some highly intelligent beings are doing things differently, reversing time鈥檚 flow in their locality. Perhaps manipulating the interactions of light and matter is the best way to establish intragalactic communications over 60,000 light years, for example. Or, of course, it may just be fun. 鈥淢ore advanced civilisations might find this amusing,鈥 Hartle says. 鈥淲ho knows?鈥