In his book Explaining the Brain (2007), Carl Craver argues that there are 3 basic kinds of experiment in neuroscience: interference experiments, stimulation experiments, and activation experiments. The first two kinds of experiment are bottom-up, involving direct interventions on the components of neural mechanism. The third kind of experiment is top-down. According to Craver, these three forms of experiment are used to help neuroscientists discover neural mechanisms. Neural mechanisms are composed of causal processes whose joint function explains a target phenomenon. For example, the mechanism of the action potential is composed of causal processes with parts including ion gradients and ion channels. Finding explanations for phenomena in neuroscience involves determining which processes comprise the mechanism of the phenomenon. The three kinds of experiment in Craver's taxonomy make possible those deteriminations.
Interference and stimulation experiments are bottom-up in the sense that they are attempts to alter the state of a phenomenon to be explained (explanandum) by interfering with component processes of its mechanism. Lesion-studies are the paradigmatic instance of inference experiments. Other instances might include transcranial magnetic stimulation (TMS), genetic knockout, and receptor blockers.
Stimulation experiments are also bottom-up experiments. As the name suggests, these combine the stimulation of mechanism components with the measurement of a dependent variable. Here, microstimulation studies are the paradigm case.
Activation experiments are top-down experiments. That is to say, they involve interventions on a target phenomenon by going through "the normal causal pathway" that affects that target. To illustrate what he means by an activation experiment, Craver writes:
"There are several common varieties of activation experiment at all levels in neuroscience. In PET and fMRI studies, one activates a cognitive system by engaging the experimental subject in some task while monitoring the brain for markers of activity, such as blood flow or changes in oxygenation... In single- and mutli- unit recording experiments, one engages the subject in a task while recording the electrical activity in neurons. In other studies, researchers monitor the production of proteins, or the activation of immediate early genes such as c-fos and c-jun. The experiments leading up to Hodgkin and Huxley’s model of the action potential involved generating action potentials and monitoring single ionic currents while the neuron spiked..."
(Craver 2007, 151)
Compare these categories of experiment with David Sweatt's system and notice that Craver does not include the "determine" class of experiment that Sweatt offers. Nor does Sweatt offer considerations regarding the top-down or bottom-up nature of experiments. Also notice that Craver's class of activation experiments assumes that some form of manipulation is being performed by the experimenter on the neural system. The manipulation might just be a psychological task that is to be performed while measures of neural activity are taken. Or, the manipulation might be some form of stimulating input, such as in the Hodgkin and Huxley example, so long as that input mimics the normal input to the mechanism (in this case, of the action potential). There are no purely observational forms of experiment in Craver's list.