Instrumental learning

The field concerned with possible brain substrates of instrumental learning is vast; there have been many thousands of studies over the 75 years since Karl Lashley began his search for the engram. A wide variety of different tasks have been used, ranging from one-trial passive avoidance to the operant procedures developed by B. F. Skinner at Harvard to maze learning and puzzle boxes.

The basic distinction between classical and instrumental conditioning is that in the latter, the animal or human controls the outcome. In instrumental avoidance learning, say a rat learning to press a lever in response to a tone CS, pressing the lever prior to the occurrence of the paw shock US prevents the US from occurring (in classical conditioning the shock US always occurs, regardless of what the animal does). In reward learning, for example, the rat learns to press a lever to obtain a food reward. The process of operant conditioning, developed by Skinner, simply involves the animal learning to make a response to obtain a reward. Skinner varied the schedules of reward and showed this had profoundly important effects on behavior. Indeed, operant techniques are very effective in dealing with severely disturbed patients.

Beginning with the extraordinary discovery by James Olds that animals would self-stimulate the reward circuit in the brain, we have learned a good deal about this circuitry (Kelley, 2004; Schultz, 2000). In brief, a system called the medial forebrain bundle projects dopamine containing neuron axons from the midbrain to forebrain structures, particularly the nucleus accumbens, the striatum and the prefrontal cortex. This system is activated by all types of rewarding stimuli, from food and water to sex. Importantly, this circuit, particularly the accumbens nucleus, is activated by all drugs of addiction. These drugs cause release of dopamine in the accumbens. Analysis of this reward-memory circuitry is a major field of research today. The dopamine projection to the striatum is of course essentially involved in Parkinson's disease and the projections to the prefrontal cortex and other higher brain regions are thought to be critically involved in schizophrenia.

Memory consolidation

The most interesting aspect of instrumental learning is the consolidation of memory. The consolidation story has two origins. In the 1940s Carl Duncan, working at Northwestern University, first made use of electroconvulsive shock (ECS) to impair memory (see Thompson, 2000). He trained rats in an instrumental avoidance task. The animals were on one side of a shuttle box, a box with a grid floor, two compartments and a connecting alley. When a light came on they had 10 s to cross to the other compartment or receive a foot shock from the grid floor. They were given one trial a day for 18 days. Control rats quickly learned the task, avoiding the shock on all but the first few days. Duncan ran a number of groups of experimental rats that received ECS (delivered through ear clips) at intervals ranging from 20 s to 14 h after each day's trial. The results were striking. Animals receiving ECS 20 s after each learning trial learned nothing at all. As the time between learning trials and ECS increased, the animals learned better and better, showing no memory impairment if the ECS came an hour or more after the training trial.

Duncan's result paralleled work in the field of psychiatry where patients with various forms of mental illness were given ECS treatments. ECS induces retrograde amnesia: events just prior to the ECS are forgotten. It has a gradient, the older the memory the better it is retained. But the gradient can be long. After a series of ECS treatments, patients may not be able to remember any of their experiences for a period of a year or more. Fortunately, most of these memories usually return, although the events immediately surrounding the ECS are usually not remembered. As with humans, so with rats.

Duncan's experiment began a large field of research. A number of possible explanations for the memory impairment were explored. Among the possibilities that were ruled out were that the ECS was strongly aversive (conditioned fear); that the ECS became conditioned to the apparatus (context-conditioned fear); that the body seizures were necessary. The fact that the ECS memory impairment also occurred when the ECS was given to anesthetized animals and humans seemed to rule out these possibilities. James McGaugh and his co-workers at the University of California at Irvine showed that the critical memory impairment could be obtained by disruptive electrical stimulation of the amygdala; seizures of the entire brain are not necessary.

The other origin of the memory consolidation story occurred early in the century in independent studies by Karl Lashley and Clark Hull, who showed that administration of strychnine or caffeine markedly improved maze-learning performance. Since they gave these substances before training, the effects could be more on the animals' performance than on memory. But McGaugh and others showed that the same memory facilitation occurred if the drugs were given shortly after training rather than before training. Possible rewarding effects of the drugs were also ruled out (see McGaugh, 2000).

Most recent work on memory facilitation has used simple one-trial learning procedures. Passive avoidance is a favorite. The animal is placed in a lighted compartment and allowed to step into a dark compartment (rats like the dark). But the grid in the dark compartment is electrified. After the animal receives a shock, it is removed. The next day it is placed in the lighted compartment and the time before it goes into the dark is measured: the longer the time, the better memory is presumed to be. This test by itself can be misinterpreted. For example, a sedative drug like a barbiturate that makes the animal inactive would produce a spurious memory. But other tests are also used, for example, active avoidance, the test Duncan used in his ECS study. Tasks involving food reward have also been used.

The bottom line in this work is that a wide range of drugs given after the learning experience can facilitate or impair subsequent memory performance in all these tasks, depending on the type of drug and the dose used. Earlier, it was thought that both ECS impairment and drug facilitation or impairment of memory acted on a specific brain process of consolidation, for example, circulating electrical activity in the brain that gradually stamped in memories. If this is so, then there ought to be a gradient of consolidation, a relatively fixed time period. However, there is no gradient, or rather there are many gradients, depending very much on the details of the procedure used in a particular experiment. This and other problems with the simple consolidation notion have led scientists to stress modulation rather than consolidation. Most workers in the field believe that ECS or drug administration modulates how well recent memories are stored in long-term memory.

Epinephrine (E) is among the most effective substances for memory facilitation, and it is of course an autonomic neurotransmitter and a critical hormone, released along with norepinephrine by the adrenal medulla in response to stress. In other words, in the real world we and other mammals tend to remember best those experiences that occur at times of arousal and moderate stress. This has been termed the "flashbulb" phenomenon - older readers will remember where they were and what they were doing when they learned that President Kennedy had been assassinated; younger readers will remember where they were when they experienced September 11, 2001 (actually, such memories are somewhat less than perfect). But we do remember best those events associated with a state of moderate arousal and stress.

0 0

Post a comment