In a nutshell, spike trains are the language in which the external world is encoded into our brains.
To understand spike trains, let us first take a look at spikes. Our brains have about a hundred billion neurons that fire signals to communicate with each other all the time. These signals are electrochemical in nature, and travel from the cell body of a neuron through its transport stalk or the axon, to the next neuron – similar to passing the baton in a relay race. Every such firing signal is referred to as a spike, or an action potential. Spikes are produced in response to stimuli or spontaneously, and each spike typically lasts for 1 millisecond.
A spike train is simply a combinatorial sequence of spikes and silences. A popular way to think of spike trains is as a digital sequence of information: 1 for a spike, and 0 for no spike. For example, an encoded spike train structure could look like 001111101101. The first two 0s represent the latency time between the stimulus presentation and the first spike. Spike trains can be induced by physical sensory stimuli such as vision, touch, smell or sound; or they can be generated by abstract stimuli such as cognitive stimulation by evoking memory.
The duration and structure of a spike train generally depends upon the intensity and duration of the stimulus. Spike trains often last as long as the stimulus is present. However, some neurons have special electrical properties, wherein they generate a sustained response to a very short stimulus. In the case of these neurons, greater stimulus intensities are more likely to generate longer spike trains.
Therefore, Justin, to answer your question, yes, constant stimuli can generate spike trains, however, some types of neurons require only brief stimuli to produce long spike trains.
Finally, I want to add that spike trains are of particular interest to neuroscientists. Here is a figure showing spike trains from thirty neurons over 4000 ms. One can ask what is the information encoded in these spike trains, or how is it decoded by neurons? Understanding spike train stimulus encoding comes in the field of study called neuronal coding – it’s a window into how neural responses embody the external world (for more see this introductory textbook). But decoding spike trains is a really complex problem. I said before that we could think of spike trains like a digital sequence of information. Now, this may give the impression that decoding a spike train would be a pretty simple affair. Computers decode digital sequences all the time!
But, there’s a catch with thinking of a spike train like a pure digital output. Neurons have a certain minimum activation threshold and they spike if the stimulus intensity is above the threshold. If a constant stimulus is presented, a spike train will be generated; however, the threshold of activation will increase over time. This sensory adaptation is the result of many biological reasons, classically including synaptic desensitization, a process wherein the synapse (the chemical connection between two neurons) is less responsive to the constant stimulus. This will lead to a reduction in spikes associated with the stimulus, eventually falling to zero. This is useful so the brain does not get overloaded with information that is not changing in the environment. You might have experienced this, for example, you stop detecting a deodorant soon after it is applied, although people can detect it as you pass them. Another example of this is our adaptation to white noise.
So every spike train encodes not only the properties of the generating stimulus, but also the state of the neural network connected to the spiking neuron. But how does the brain keep track of what spikes are due to sensory information, and which pauses are because synapses have desensitized? That is a question many neuroscientists are hoping, one day, to answer.