My suspicion is that the people who know about neuroscience read the title of this and said: "Wow, Jake, there's a shocker. Tell us something we didn't know." Everyone else probably said: "Guh?" Therefore, I should probably explain why I think this finding is cool.

LTP or Long Term Potentiation is an experimental paradigm that is believed to simulate learning in vivo. In the paradigm, an exciting electrode and a recording electrode are placed into the brain of the animal, usually into an area called the hippocampus. The electrodes are position such that the action of the excitation electrode activates the neurons near the recording electrode through a single synapse.

During the experiment, you activate the excitation electrode using what is called High Frequency Stimulation (HFS). HFS is neuroscience code for blasting the bejesus out of that synapse with electricity. The recording electrode is measured using test pulse from the excitation electrode before and after the experiment.

The phenomenon of LTP refers to the observation that after you blast the bejesus out of this synapse, the synapse becomes more effective at activating the next cell over -- near to the recording electrode -- than it was before. This is what is called potentiation at the synapse. Furthermore, potentiation is long lasting, even to the point of months -- hence the long-term part.

It has been the model for many years that LTP across many synapse in the brain -- forming a network -- is the mechanism for learning. However, I emphasize that this is an experimental paradigm because determining whether or not this is the real mechanism of learning is a tricky business. See, ideally, you want to use a physiological activation of the neurons and record potentiation changes from that. Unfortunately this is rather difficult because A) the changes from physiological stimulation are much smaller and B) the neurons form a network and you are never quite sure which synapse is going to potentiated.

Whitlock et al. publishing in Science have solved this problem and verified that LTP caused by learning -- a physiological activation -- causes measurable LTP in the hippocampus of rats. This finding is huge because it confirms how we thought memory works but could never prove.

The researchers solve one of the technical hurdles to measuring LTP using an avoidance learning paradigm. They place an array of electrodes into the hippocampus of rats. Then they place them in a box with a black area and a white area. When allowed to cross into the black area, a foot shock is administered to the rats. It has been shown from a variety of work -- and I think common sense supports -- that rats learn not to go in the black area after a single trial of this paradigm. Since learning is so rapid, it is reasonable to assume that the memory trace only requires one trial as well, and this is what their data shows.

To measure the LTP, the researchers examine many synapses across the hippocampus both before and after learning using the array of electrodes. What they are measuring in this case is what we call the synaptic efficiency -- the ability of a presynaptic neuron to activate a post-synaptic neuron. LTP is detected as an increase in synaptic efficiency.

So here is a figure of their data (click to enlarge):

In the figure, Part A shows the anatomical placement of the array of electrodes in the rat hippocampus. Part B shows what the data for synaptic efficiency looks like. The traces are measures of the depolarization of the postsynaptic neuron on activation by the activating electrode (in this case just being used to measure rather than to potentiate). A more efficient synapse would have a deeper slope to this trace because it would be depolarized more (for those who would catch this, the depolarization of the cell is shown going downwards because the recording electrode is extracellular).

The rest of Part B shows the synaptic efficiencies represented as what are called Excitatory Post-Synaptic Potentials (EPSPs). The number on y-axis is a measure of the slope of the traces above (slope is often used as a measure of the depth of the trace) for repeated time points after the learning paradigm was administered.

You can see that the red-orange is the trace of a synapse that was potentiated. After the learning paradigm the synaptic efficiency has gone up as is shown by the increasing slope of the EPSPs that persists over time. On the other hand, the green-blue is the trace of a synapse that was unaffected.

The blocks of color on the right show the behavior of synapses measured across all the electrodes in the hippocampus with the level of potentiation indicated by color. You can see that after the avoidance learning paradigm many but not all of the synapses in the hippocampus show potentiation. This does not occur either when the rat was just shocked, or when it was allowed to wander around the box without being shocked.

I think this experiment is particularly elegant in verifying that our theory of LTP based learning is correct.