The historical unraveling of the memory systems present in the mammalian brain has been an extensive journey of scientific discovery. Through the work of countless individuals, major hypotheses have been proposed which have led to the current view of how memory is stored and retrieved. Though many uncertainties and controversies still exist, the general mode of memory persists, with substantive research corroborating the principal findings. Research thus far has concluded that there are two major forms of memory: declarative (explicit) and nondeclarative (implicit). These forms of memory are defined by the nature of what is remembered, but memory can also be categorized based on the time in which it remains effective. These classes of memory can be divided into ‘immediate’ memory, ‘working’ or ‘short-term’ memory, and ‘long-term’ memory. A further analysis of how these separate forms of memory exist in the mammalian brain, which neuroanatomical substrates are important in their actions, as well as the general characteristics of each type of system will be conducted. But first, it is important to examine the historical events and characters that have led to the present understanding of these distinct memory systems.
The first discipline to examine memory and question how learning occurs and how memories are stored was philosophy. As far back as the B.C. time of Socrates, philosophers were questioning how humans learn new information and how they store it as memories. Using methods of conscious introspection, logical analysis, and argument, they sought to explain how memory worked. However, these non-experimental methods did not lead to agreeable conclusions. Later, a French philosopher named Maine de Biran was the first to have the idea that memory could be separated into different systems. His notion was that memory could be distinguished into three distinct forms based on habits: representative memory, mechanical memory, and sensitive memory. Representative memory was characterized as the recollection of ideas, mechanical memory dealt with the habitual repetition of a movement, and sensitive memory referred to the habitual generation of a feeling without recalling the ideas behind it.1 De Biran remarkably theorized these three distinct forms without any actual experimentation or consideration of the anatomy of the brain. His theories, although not exactly correct, highly resemble the modern day accepted forms of memory systems. Albeit, his theories were not well accepted until much more recently, when the idea of separate memory systems was found to be strongly supported by modern cognitive neuroscience. Others, like Franz Joseph Gall, also had ideas that there were different types of memory systems. Gall was the founder of phrenology, the idea that you could feel the bumps on the head of a person and get an accurate idea about their personality and other characteristics, and theorized that there were different types of memory localized in certain spots of the cortex. He was wrong mostly, but correct about different types of memories localized in different areas.
By the nineteenth century, trying to explain the complex workings of memory through philosophical methods was replaced by the empirical methodology of the newly-founded discipline of psychology. Experimental psychologists like Hermann Ebbinghaus showed that memories have different life spans, while William James further explained the fundamental difference between short-term and long-term memories, as well as drew the distinction between declarative memory and procedural memories. At the same time, psychiatrist Sergei Korsakoff was the first to document a memory disorder that resulted in human amnesia. In the twentieth century, another prominent psychologist figure named Karl Lashley sought to ‘localize the engram’, or find the one, or couple, localized regions of memory. He did this by testing the maze-running ability of mice when different sections of cortex were surgically removed. He found that mice were still able to complete the maze no matter the lesion, which led him to conclude that memory was not localized to a single region and was widely distributed throughout the entire brain. This ‘mass action view’ of memory stuck with psychologists and was the major theory of memory for many decades. With hindsight, the choosing of a maze-running task to examine memory was not ideal, as maze-running involves many brain systems, from vision and olfaction to spatial sense and others, and cannot be pinned to just general root memory. As it is now known, Lashley’s influential theory on ‘whole brain’ memory was partially correct in the aspect that memory is widely distributed. However, he assumed that memory was comprised of one unitary system that was widespread and stored all over the brain. The view by Lashley that memory was not localized held sway up until the 1950’s, when a new style of memory research came into play. Interestingly, Lashley influenced psychologists to believe that with brain injury other areas of brain tissue would be able to substitute for the function of the affected region.2 This view by Lashley will come to be challenged by the work that was soon begun by researchers with patients who have amnesia or other memory disorders.
Recent researchers circled back to the work of Korsakoff and began investigating disorders causing memory impairment, often brought on by lesions in certain regions of the brain. It wasn’t until researchers realized that persons with memory disorders could elucidate the structure and organization of normal memory that huge leaps in the understanding of memory systems occurred. Psychologists and researchers began looking extensively at the cause and function of amnesic patients to help explain how memory worked in the brain.
Canadian psychologist Brenda Milner’s study of the amnesic patient H.M. led to many revolutionary findings that formed some of today’s understandings of the memory systems. Brain injury as a young boy plagued H.M with frequent seizures, which led to him receiving an experimental treatment. The treatment included the bilateral removal of the inner surface of the temporal lobe, including the hippocampus. This treated his epilepsy, but left him with large-scale memory loss that compromised his ability to form new long-term memories. From this led to a comprehensive 50-year study of his memory condition. H.M.’s most drastic deficit was that he appeared to forget events as soon as they happened. Although he could retain new information as long as his attention was not diverted from it, but as soon as a new task was begun, that previous information was forgotten. Milner was able to extract four novel principles on memory from her study of H.M.. First, she alleged that the ability to acquire new memories is a distinct cerebral function, localized to the medial portion of the temporal lobes of the brain, and separable from other perceptual and cognitive abilities. This was evident as H.M. was without the medial part of his temporal lobe and could not form new long-term memories. Second, she hypothesized that the medial temporal lobes are not required for immediate memory. H.M. could still retain information for a short time after learning, and carry on a brief conversation. Third, she said that the medial temporal lobe and the hippocampus cannot be the ultimate storage sites of long-term memories, since H.M. still had complete recollection of events from his childhood and time before his experimental treatment. Lastly, Milner found that H.M. could still learn a type of knowledge allowing for perception and intellectual functions. In an experiment, it was discovered that H.M. could learn to trace the outline of a star in a mirror, and with each day of practice there was improvement, even without remembering previously completely the task. It seemed as if there was a separate type of memory that did not originate in the medial temporal lobes.3
In a critical evaluation of Milner’s conclusions, it would be important to note that H.M. had such an extensive lesion of the medial temporal lobe that it makes it difficult to attribute these theoretical functions to the region as a whole. It could be very likely that only small regions of the medial temporal lobe (MTL) play a role in each of H.M.’s memory impairments, and it was hasty of Milner to attribute the acquiring of new memories to the entire region. Also, in evaluating the findings now, it could be possible that formation of new episodic memories could follow a neural pathway involving more than one region of the brain, with the MTL acting as intermediate location in the pathway of formation of new episodic memories from short-term ones. Disrupting the function of these substrates could in essence ‘cut the cord’ in the pathway, disrupting the pathway for coding the episodic and semantic information. With this in mind, it could elucidate that the substrates composing the MTL are important for the formation of new episodic memories, but not the main center for this memory system. Further evaluation of other subjects with varying degrees of memory impairments caused by different lesions could reveal more information about the memory system proposed by Milner.
Finding subjects with amnesia to study for multiple years, as well as being able to examine their brains post-mortem for damage makes this realm of memory study difficult. Over the years, there have been several more amnesiac patients that have been well-studied and can offer clues into the brain’s systems of memory along with H.M. Two other well-studied amnesiacs, R.B. and G.D., developed amnesia following ischemic episodes involving a loss of the blood supply to the brain. Like H.M., their general intelligence and cognitive abilities were unimpaired, but both experienced trouble forming new memories. A post-mortem evaluation of their brains showed extensive bilateral damage to the CA1 field of the hippocampus, with no sign of damage elsewhere, fortifying the theory that the hippocampus is important in the acquisition of new episodic memories. It also highlighted the idea that damage to small subregions of the hippocampus can cause extensive memory impairment. R.B. and G.D.’s impairment was only moderate in comparison to H.M. though, which could signify that other regions of the hippocampus play roles in memory.
Two other subjects, L.M. and W.H. displayed more severe impairment than R.B. and G.D. in terms of anterograde amnesia, and also exhibited extensive retrograde amnesia. After post-mortem evaluations, it was shown that both had extensive damage to the CA1 field of the hippocampus, as well as the CA2 and CA3 fields, the related area called the dentate gyrus, the outer edge of the hippocampus called the subiculum, and a neighboring area called the entorhinal cortex. Altogether, it seems conclusive that the hippocampus and MTL play a large role in memory, as more damage to the hippocampus and the surrounding MTL substrates elicited increasing memory impairment. It was evident that one or all of the additional regions affecting R.B. and G.D. (CA2, CA3, dentate gyrus, subiculum) played a role in causing retrograde amnesia, posing the idea that one of these regions is responsible for the housing of past declarative memories, or at least a major substrate in the pathway responsible for storage and/or retrieval of past declarative memories. However, even with the data presented by these cases, it is unclear the specificities of memory that each individual region of the MTL plays in the declarative memory system. And again, it could still be possible that the MTL acts as a sort of the ‘hub’ or the main location along the episodic memory formation or retrieval pathway, but still could have other brain regions operating in this memory system. Clearly, it is evident that another form of research needs to be conducted where the individual structures and connections of the MTL and other regions can be examined to determine the complexity of the declarative memory system.5
With the theories provided by the study of these amnesic patients, it is clear now that memory does in fact have highly localized regions that come into play with certain types of memory. Now that the basic human model of declarative memory was understood from amnestic patients, these specific regions that played a role in declarative memory could be examined in more detail using animal models. An animal model was constructed with the nonhuman primate; one of humans’ closely related species in terms of brain function and structure. With the animal model, the specific structures of the MTL that are essential to declarative memory came to light. Although it’s evident that animals cannot express declarative memories in the form of conscious remembering the same way that humans can, researchers have managed to determine ways in which to test declarative memory in these animals. They look at declarative memory’s other operating characteristics by examining the kind of information that is processed in certain tasks given to the monkeys. The first task is the ‘delayed nonmatching-to-sample’ task and entails the animal to simply recognize an object as familiar when first being presented with the object as a sample, and then after a delay, spanning from seconds to minutes, presenting the same object along with a new, novel object in which they’ve never been exposed to. If the animal subject chooses the new object, signifying that they recognize that the other object is familiar, they get rewarded with food. In the second task, the subject is presented with two simple, easily distinguishable objects. One of the two objects is designated the ‘correct’ object, and if the animal chooses the correct object they received food as reward. A normal monkey took between ten and twenty trials to learn the correct object.
Using animal models provided the unique opportunity to surgically induce lesions at specific chosen regions of the brain and run them on the same tasks, giving remarkable insight to the physiological function of certain regions of the brain. As you might imagine, this is not possible, for ethical and other reasons, in human subjects. The only memory impaired human subject available to obtain information regarding memory systems either has lesions resulting from accidental events, conditions like cerebral anoxia, or simple genetics resulting in atypical brain conditions. Memory impairment diseases like alzheimers have become the subject of much research lately in terms of its effects on memory in regard to the declination of healthy brain tissue.
It took a while for these views on memory systems provided by the study of amnesic patients to gain ground, as many at the time thought amnesia was merely caused by a retrieval deficit. Additional experiments on amnesiacs were conducted to broadly examine the information that could be gathered on different memory systems of the brain. In well-structured experiments where three-letter word stems were presented with instructions that told the subject to use this word stem to form the first word that comes to mind, amnesiacs were able to perform just as well as the control group. Tasks using this sort of instruction became known as ‘priming’, and showed that it was a distinct form of memory that was unimpaired when MTL regions were damaged. It can therefore be deduced that procedural tasks involving priming involve brain regions outside of the main memory MTL region, possibly in the cortex. If the task instructions were phrased in such a way to mimic conventional memory, like telling the subject to use the word stem as a cue to retrieve a word that was previously presented, the amnesiacs performed much worse. This was significant, as amnesiacs often experienced the inability to encode and store previously presented words. Furthermore, similar to H.M.’s mirror star-drawing task, other tasks of the sort (mirror reading, resolving stereoscopic images, cognitive skill learning, artificial grammar learning, and category learning) were conducted with other amnesiacs. Although they had poor memory of the tasks they had just completed, over time they experienced improvement in the specific task. These tasks elucidated a different form of memory using skill-like abilities that was also unimpaired in amnesic subjects with various MTL lesions. Data from these experiments showed a distinction between declarative and nondeclarative procedural knowledge memory systems. However, the broad division of memory between those two systems were challenged, as influential experiments showed many more types of memory present in the brain.6
Experiments examining associative learning used the method of eyeblink conditioning, in which a simple associative task using a conditioned stimulus like a light or a tone, paired with an unconditioned stimulus like a peri-orbital shock, elicited a reflexive eyeblink. Using variants of trace and delay conditioning, the unconditioned shock stimulus became paired with the conditioned light or tone, and after several trials, a reflexive eyeblink was observed even when the light or tone was presented on its own without the shock. Researchers tested this method with animals in which all brain tissue above the thalamus or midbrain was removed and the effect was still present.7 It was concluded that the cerebellum was essential to this type of conditioned learning. The effect was also shown to still be present in amnesiac patients, eliciting that again there was another form of memory centered outside the standard declarative hippocampal area regions.8
The data compounded from all these experiments could be shown as memory systems distinguishable by the figure below (Figure 1). The broad division of memory remained distinguishable between declarative and nondeclarative systems, but further evaluation led to the umbrella of individual systems that contributed to the two broad divisions. Overall, it has been shown with forms of evidence that the hippocampus and the MTL substrates are responsible for forming new declarative memories, although the storage of those memories happens in the frontal lobe and elsewhere in the cortex. The nondeclarative memory systems have a wider array of inputs into the generalized broad system, as many parts are seen to contribute to the overall encoding, storage, and retrieval of nondeclarative memories, all outside of the general MTL localization. If anything is to be realized from the experiments conducted thus far on amnesiacs and animal models, it is that memory as a whole is not localized to a single region of the brain as some early philosophers reasoned, but rather followed a combination of Lashley’s view of ‘whole-brain’ memory in combination with Gall’s theory that different forms of memory were localized in different regions. As a conclusive interpretation of memory in the brain, it can be stated that memory systems are localized to different parts of the brain, but interconnect in ways to form memory pathways that still remain not fully understood.
Do non-human animals even have episodic memory? It is even possible that while humans learn some tasks, like with visual pattern discrimination, in a declarative manner, other non-human animals learn the same tasks nondeclaratively. If animal models are not learning and using their memory systems in the same way as humans, how can they be considered representative as to how humans process memory? Humans often approach tasks as simple problems of memorization, while monkeys and other animals gradually learn the visual pattern discrimination over lots of trials in a way that resembles skill learning.
LONG-TERM STORAGE IN NEOCORTEX
Frontal cortex has the function of holding information for impending action
The view stemming from Milner’s study of H.M. suggested that regions outside of the MTL are critical for the long-term storage of declarative memories.
Direct damage to these surrounding MTL areas impairs memory even more severely than damage to the hippocampus proper. Important for path integration between working memory and long-term memory, includes the proposal that the path integrator is located in these structures.
Amnesic patients were used in a task where they were lead on a path, blindfolded, of no longer than 15 meters and included up to three turns. At the end of the path, the subjects were asked to point in the direction of their starting location. If the subjects were told to actively keep the path in their mind, they were as accurate as the control group. However, when the researchers increased the demand of long-term memory by waiting a few minutes, the amnesiac subjects performed significantly worse. Two patients with the most severe damage (E.P. and G.P.) were unable to remember anything of what they had been doing several minutes after the task. Further proves that MTL regions are imperative for long-term memory. Immediate memory and working memory are still intact with damaged MTL substrates.
Flexibility of declarative memory and relative inflexibility of nondeclarative memory, shown in a study of spatial learning and memory in a rat (page 109)
Now biology is coming into play, as molecular components are being combined with psychology to explain the inner-workings of the brain and its complex memory systems
There are many forms of memory, different brain structures carry out specific jobs, memory is encoded in individual nerve cells and depends on the strength of their interconnections
2016-9-20-1474342870