How does light 'choose' between wave and particle behaviour? Light exhibits wave behaviour in phenomenon such as interference but particle behaviour in the photoelectric effect. How does light 'choose' where to be a wave and where to be a particle?
 A: 
How does light 'choose' where to be a wave and where to be a particle?

It doesn't. It always behaves as a wave (obeying the principle of superposition), and it always behaves as a particle (particle number being quantized).
It sounds like you may have been influenced by someone who told you that light behaves like a particle in some experiments, and like a wave in others. That's false.
A: Light always behaves as a particle and waves .So there is no particular time when it can behaves like a particle but not wave and vice versa .Thus light carries both of these two nature(particle and wave) along with it (light)for all time.
Actually why is my thinking is like light has two of these nature for all time?
Reasons for considering it as *wave * 


*

*James Clerk Maxwell proved that speed of the electromagnetic waves in free space is the same as the speed of light c=2.998 ×10^8m/s. Thereby he concluded that light consists of electromagnetic waves .So it has wave nature for all time.

*Light also shows *interference *,*diffraction *,*polarization *.
And all of these property show that light has wave nature and it is for all time.


Reasons for considering light as particle


*

*Max plank in 1900 introduced concept of quantum of energy .The energy exchange  between radiation and surroundings is in *discrete or *quantised * form i.e energy exchange between electromagnetic waves is integer multiple of (plank constant *frequency of wave ).As light consists of electromagnetic waves then light is made of discrete bits of energy i.e photons with energy (plank constant *frequency of light).
Thereby photoelectric effect ,Compton effect all show the particle nature of light.


So radiation (electromagnetic waves) exhibit particle nature conversely material particles display wave like nature introduced by de Broglie(a moving particle has wave properties associated with it).This is confirmed experimentally by Davisson and Germer they proved that interference which is a property of waves can be obtained by with material particles such as electron.
So as radiation exhibit particle like nature but particles also exhibit 
wave- like behavior .Thereby light always shows two type of natures both as waves and particles .
A: The fundamental experiment showing the apparent contradiction is Young's double slit: How can particle characteristics be transmitted when there is only an interfering wave between the point A of emission and point B of absorption?
However, for photons in vacuum (moving at c) there is a simple answer: The spacetime interval between A and B is empty, it is zero! That means that both points A and B are adjacent. A and B may be represented by mass particles (electrons etc.) which are exchanging a momentum. The transmission is direct, without need of any intermediate particle. 
In contrast, a spacetime interval cannot be observed by observers. If a light ray is transmitted from Sun to Earth, nobody will see that A (Sun) and B (Earth) are adjacent. Instead, they will observe a space distance of eight light minutes and a time interval of eight minutes, even if the spacetime interval is zero. In this situation, the light wave takes the role of a sort of "placeholder": Light waves are observed to propagate at c (according to the second postulate of special relativity), but this is mere observation.
In short, the particle characteristics may be transmitted without any photon because the spacetime interval is zero. The wave characteristics (including the propagation at c) are observation only.
By the way, for light propagating at a lower speed than c (e.g. light moving through a medium), we need quantum mechanics for the answer. 
A: The light doesn't chose. You as an experimenter chose which observable you want to measure, and thus which operator you use. Such measurement will result in a wavefunction collapsing into one of the eigenstates of this operator.
E.g. a positiin operator will give you a position, i.e. particle.
A momentum operator will give you a momentum, i.e. wavelike object.
A: Light is assumed to be particle or wave to describe the observed phenomenon. Light is what it is and doesn't choose. However, one can consider light to be a particle when the interaction is with matter & as a wave when it interacts with itself. 
A: It doesn’t.
Light, as you may know, is made up of photons, and photons are particles. At the same time, all particles are waves. Yes, that's right—particles are not points or spheres or anything else really solid moving around, but waves, which interact with each other in various ways.
Now, these waves may sometimes behave indistinguishably from points moving around in space, and may therefore be approximated as such—what we like to refer to as point particles—but they are fundamentally still always waves.
So, a photon is always both a wave and a particle; these are not mutually exclusive (in fact, as I have explained now, particles are a subset of waves). This is what is known as the wave–particle duality.
A: It doesn't is the short answer, as it is really neither.  Basically, waves and particles are things you see on the classical scale, they model behaviour on macroscopic scales.
Objects don't behave in a classical way any more when we shrink down to very small scales and they obey very different dynamics which seem strange to us because we do not live on that scale and are only able to access things on that scale in quite a dim way via theory and experiment.  It's kind of common sense when you think about it, just stop expecting that things will behave in a classical way when you going down to those length scales.
Also forgot the wave-particle duality, as it does not even make sense.  It's like a blind person touching an elephant on the trunk and then touching it again on the leg and saying that the elephant follows trunk-leg duality.
A: 
How does light 'choose' where to be a wave and where to be a particle?

It doesn't, YOU do. That's really the entire "weirdness" of quantum right there.
It's not entirely crazy. If you measure a car with a scale it will tell you it's 1200 kg, and if you measure it with a spectrometer it will tell you its red. This is perfectly natural.
The thing that makes quantum weird is that you can measure the same thing twice and get two different answers. More weird, some of those measurements are linked to each other so if you measure one the other changes.
Its as if you measured the weight of your car and the length changed. And then you measured the length and the weight changed. This precise thing happens in quantum, for instance, the position and momentum of a particle are linked in this way
In any event, the wave-or-particle nature is entirely up to you. Which nature you see is based on the experiment you use, not the photon itself.
The idea that something you do has this "radical" effect on the outcome is what drives everyone mad in QM.
A: In fact, light is not really a wave or a particle. It is what it is; it's this strange thing that we model as a wave or a particle in order to make sense of its behaviour, depending on the scenario of interest.
At the end of the day, it's the same story with all theories in physics. Planets don't "choose" to follow Newtonian mechanics or general relativity. Instead, we can model their motion as Newtonian if we want to calculate something like where Mars will be in 2 weeks, but need to use general relativity if we want to explain why the atomic clock on a satellite runs slow compared to one on the ground.
Light doesn't "choose" to be a wave or a particle. Instead, we model it as a wave when we want to explain (or calculate) interference, but need to model it as a particle when we want to explain (or calculate) the photoelectric effect.
A: Light always behaves as a wave. Particles can be thought of as a combination of waves, a wave packet. What determines which behavior you'll get is the length scale of the system the light is interacting with and the wavelength of the impinging light. 
Suppose you have a series of Bowling Balls suspended in the air making a wall, but with spaces between. Shoot a bunch of 2mm diameter ball bearings (BBs) at the bowling balls. Some will pass through the wall, hitting no bowling balls. Some will hit a ball and bounce in some direction. 
Pay close attention to where the BBs come from and their initial speed and where they end up, and you will be able to tell the shape, size, and position of the bowling balls. 
Reverse the problem. Have a bunch of suspended BBs in the air and shoot bowling balls at them. Each bowling ball will hit multiple BBs. The outgoing trajectory won't tell you much about the BBs. 
Strike a tuning fork, it vibrates, making a characteristic sound. Play that note at high volume, you can set the tuning fork to vibrating. It doesn't start to vibrate at just any wavelength. 
To a particle, you can associate the De Broglie wavelength, $h/p$, where $p$ is the momentum. The higher the momentum, the lower the wavelength, the more particle like. Whereas macroscopic but small openings can be used with light in the double slit experiment, you need electron crystallography to demonstrate the same effects with electrons: Electron diffraction
If you have a large wavelength of light, it will interact with multiple particles of a system depending on that system. If the wavelength is sufficiently small, and so the energy sufficiently high, it can interact with a single electron instead of multiple electrons, imparting all its energy to an electron and resulting in the photoelectric effect, a particle-particle scattering effect. Change the wave length and the interaction takes on a more classical form. 
In addition to the wavelengths involved you want to pay attention to the number of photons available. The fewer the photons interacting with the system, the more quantum-like it is. The higher the density of photons, the more classical the light will behave. For a more detailed explanation on the barrier between quantum vs. classical behavior, see the intro and first chapter of Griffith's text on electromagnetism.
In short, the behavior you get depends on the De Broglie wavelength of the light/particle, how many particles are inbound and how the lengths scales of the target compare with the De Broglie wavelength. 
A: binaryfunt and Ben Crowell's answer is closest to the truth. Let me add a few things.
You get confused because you read sentences like, light behaves as a wave when it travels in space and as a particle when it interferes with matter. That is not true. In reality light has characteristics of both waves and particles at the same time. The truth is, we do not know what actually light is, or what it is really made of. 
In reality, light is not what decides or behaves different ways, it is us, the ones preparing the experiments, who decide whether we will select from the observed (both wave or particle like) nature the more interesting wave or particle like characteristics, and analyze those for certain purposes.
What we try to do, is model with our mathematical ways the real phenomenon in our physical world, light itself.
The reason you get confused is because there are two ways to really model this phenomenon:


*

*classical EM waves

*QM particles, photons
You would think they do not match, but that is not true. In reality the two methods come together perfectly, as we build up the classical EM waves from herds of QM photons.
In our currently accepted theory, the Standard Model, we talk about elementary particles, that have no internal structure, nor spatial extent, and we call them point like particles. One of these particles is the photon. We call it the quanta of the EM wave.
It is very important to see that the basic confusion lies in whether you want to analyze the wave or the particle characteristics of light in a certain experiment.
We usually model light in two ways, and that is what confuses you, these to ways are whether it travels or when it interacts with matter. This is basically a misconception, because light could be analyzed to show wave and particles characteristics both when it travels and when it interacts with matter.
It comes from a misconception to choose between travel and interaction, but for the sake of argument, lets see:


*

*travel


You can use Maxwell's original equations or use QFT to treat photons and light as excitation of the EM field, that travels through space, and this propagation is modeled in our theories as a wave. This is because this wave model is what best fits the experimental data. Now we can show that the QM or QFT field excitation model of the photon propagating through space is the same fitting the data perfectly from the experiments.
So you see light has both wave and particle characteristics even as it travels through space.


*interaction with matter


Now when a photon interacts with an atom, three things can happen:


*

*elastic scattering, the photon gives all its energy to the atom and changes angle

*inelastic scattering, the photon gives part of its energy to the atom and changes angle

*absorption, the photon gives all its energy to the absorbing atom/electron
For all three, you can see that light both shows wave and particle characteristics. It is a misconception that the experiments about interaction are about the particle characteristics alone. Even in these three cases, when light interacts with matter, it behaves as a wave, in certain cases it changes angle, goes through slits, splits into partial waves, these partial waves interact with themselves, create interference patterns. Even in the case of absorption, when the photon ceises to exist, and transforms into the kinetic energy of the electron, it shows wave nature and particle both since the absorbing atomic system and electron both have wave and particle characteristics too.
So there is no reason to really make a hard decision whether light would choose, it does not. It has both wave and particle characteristics as intrinsic, and it is up to us who prepare the experiment and analyze the data which we want to focus on.
