Researchers once considered memory to be a unified brain system, but findings over the past 30 years have changed that way of thinking: Neuroscientists now know that memory systems are housed in separate parts of the brain and recent evidence suggests they can conflict and interact in sometimes surprising ways.
During an invited address at APA's 2004 Annual Convention in Honolulu, Mark Packard, PhD, an associate psychology professor at Texas A&M University, recounted the evolution of that research. He also discussed work in his own lab, which finds that the brain's memory systems work differently during stressful or emotional times.
Past research has shown that largely separate memory systems are involved in learning different kinds of tasks, Packard said. For example, the hippocampus is key to cognitive learning, such as spatial mapping--useful for rats remembering where to find food in a maze when they are released from a new starting point--while the caudate nucleus mediates stimulus-response or habit learning--such as simply learning to turn left in a maze to find food, regardless of the starting point.
Damaging one of these respective brain areas wipes out an animal's ability to learn specific tasks. For example, hippocampal system lesions impair spatial learning, while caudate nucleus lesions impair habit learning--findings replicated in humans and other primates.
Moreover, Packard's research has shown that injecting memory-enhancing drugs into the hippocampus can heighten performance on tasks that require cognitive memory, while boosts to the caudate nucleus improve rats' performance on habit-learning tasks.
However, said Packard, rats with hippocampal lesions actually perform better than undamaged rats on some habit-learning tasks that use the caudate nucleus. The findings suggest that the hippocampal memory system sometimes interferes with learning in the caudate nucleus.
Shifts in learning
The interactions between the two memory systems become even more intriguing, said Packard, as researchers examine learning over time. He pointed to studies dating back to the 1950s, which found that, when given the choice, healthy rats initially use cognition to learn and perform tasks but eventually switch to habit learning. Packard's lab has conducted several studies to explore the neural basis of this tendency.
For example, he and his colleagues trained rats in a food-finding task by placing them at the bottom arm of a plus-sign-shaped maze to find food at the end of the left arm. Packard then probed the animals' learning by releasing them from the top arm; rats who learned the task cognitively would turn right and go to the same arm where they'd found food during training, suggesting that they had learned the food's spatial location. Conversely, habit-learners would instinctively turn to the left--the same body turn response rewarded during training. On the seventh day of training, rats showed place-learning, but by the 15th day, they had switched to habit.
His team found that damage to the hippocampus prevented rats from place-learning, but that rats with caudate damage were place-learners early on and in the later trial.
In other studies, Packard gave the rats memory-boosting injections of glutamate in either the hippocampus or the caudate nucleus early in training. After seven days, the hippocampus-injected and control animals showed place learning, but the caudate-injected rats had already shifted to response learning.
"By boosting this response learning very early in training you can essentially override the place learning that's normally going to be controlling behavior," he explained.
Moreover, after 15 days, the animals with hippocampal injections continued to show place learning--suggesting that the brain's typical switch to response learning was blocked.
Taking cognition offline
In more recent studies, Packard and his colleagues find that emotional state can influence the use of these two memory systems. The research shows that, while unstressed animals normally start out as cognitive learners and shift to habit learning only over time, stressed rats seem to learn by habit earlier.
For example, Packard and his colleagues trained rats on one of two plus-maze swimming tasks--one in which the rats must always swim to the same arm and one in which the rats must always turn left. When they then injected anxiety-inducing drugs into the rats' amygdalas, the animals did poorly on the place-learning task, but actually did better than control animals on the habit-learning task.
"Here's an example where [anxiety] is taking cognition offline," allowing habit learning to control behavior, he explained.
The findings may shed light on the role of stress and anxiety in human disorders such as compulsive drug use or obsessive-compulsive disorder--in which activating habit-learning systems may result in maladaptive behavior.
Letters to the Editor
- Send us a letter