Quantum physics' two-faced nature has been put to the test over and over again, and every time it's still damn weird.

This time physicists have gone to some lengths – literally – by splitting and merging light and then bouncing it from a satellite before testing its odd-ball properties. And, yeah, even over a stupidly long distance, it's still messing with our heads.

A team of Italian physicists has gone super-size with something called Wheeler's delayed-choice experiment in an effort to see if the process can be scaled up from a previous record distance of 144 kilometres (about 90 miles) to 3,500 kilometres (about 2,200 miles).

Not to give away the ending, but you're probably not surprised to learn the experiment's results still hold true, and the form a wave of energy takes seems to depend on how a conscious mind looks at it.

But the insanity behind it all is fascinating, so let's take a moment to dig into the history of the experiment.

Back when quantum physics was still in its infancy, the smartest minds in the world of physics were leaning ever more heavily on the mathematics of probability to explain what they observed.

As useful as the mathematics was, the consequences were profound. Some – like Albert Einstein – figured the probability stuff was temporary, and more solid laws would one day be uncovered. Niels Bohr, on the other hand, said quantum mechanics was complete and while our brains don't cope well with trying to imagine 'maybe' as a solid part of reality, well that's just tough.

This all came to a head over a series of thought (and actual) experiments to do with light – and indeed, matter of all kinds – behaving as both a wave and a particle.

Consider something called a double-slit experiment: the 'maybe' wave nature of a particle of light means if given two slits, it will pass through both of them before hitting a screen on the other side, causing a pattern as each possible path interferes with one another.

If we stick something in the way to detect which path the light 'really' took, we no longer see that pattern. The photon was revealed as a single, high speed projectile shooting through one of the two slits.

Bohr was adamant those 'maybes' collapsed into a 'really' depending on how the experiment was conducted. Einstein thought this was baloney, as it meant the reality of a particle – complete with its position, velocity, spin and so forth – didn't exist until we measured it.

To cut to the chase, numerous experiments over the years have shown Bohr to be on the money. 

But there has long been one, nagging question - what exactly constitutes an experiment? Does light cement itself as a particle or a wave the moment it interferes with the relevant equipment, or are our minds part of the whole set-up?

An American theoretical physicist John Wheeler devised a clever way to test this about 40 years ago.

Putting it simply, why not trick the light by conducting one kind of experiment that forced it to choose its reality, and just before it hits the last detector another piece of apparatus randomly forces a sneak peek?

This was the basis of a number of tests described as 'delayed choice experiments'.

In this latest version, the researchers split a pulse of laser light using Italian Space Agency's Matera Laser Ranging Observatory (MLRO), so a photon could either take a shorter path, or be sent on a slightly more convoluted detour.

The two paths merged before heading on a several thousand kilometre journey to an orbiting satellite, from where the photon bounced back to the planet's surface.

Back at the MLRO, a quantum random number generator – about as random as we can get – chose whether or not a device delayed the incoming photon.

Letting it through would mean the researchers knew which path it took; the equivalent of measuring it as a defined particle with a clear history.

Delaying the photon would make it impossible to tell whether it had arrived early or not, leaving its history unknown.

Importantly, that random decision was made well after the photon set off on its extensive journey, thousands of kilometres in the past.

The researchers found they could affect whether the photon was perceived as a wave or a particle well after it had passed through the important parts of the experiment.

It's as if the random decision to let the photon pass through made it go back in time and choose one path, while delaying it meant it still had a history of possibilities, each of which would then interfere with its detection to reveal a wave-like nature.

This doesn't completely solve the mystery of what it means for a wave to collapse into a particle, or how our minds are involved.

But even after the horse has bolted 3,500 kilometres out of the gate, it can still wait until the finish line to decide which race it ran. 

Sorry Einstein. Can't win 'em all. 

This research was published in Science Advances.