Divyan Bavan
Introduction
Memory is a critical aspect of our brains. It enables us to learn from our past experiences, integrating that information into future responses. Thus, scientists have been trying to understand this process for a long time. One of the ways we do this is through looking at patients with memory impairments. By connecting what caused their impairment to its consequences, it is possible to suggest a theory for how that process works. For memory, the most famous example of this is Patient HM. After going through bilateral removal of the medial temporal lobe, Patient HM suffered from severe amnesia. This suggested that structures within the medial temporal lobe, such as the hippocampus, are critical for forming memories. While this formed the basis for our understanding of memory, several other studies have refined this theory. This has led to the creation of a taxonomic tree for long-term memory.
The Dissociation of Long-Term Memory
Long-term memory can be split into two categories: declarative and nondeclarative. The former is concerned with explicit information, whereas the latter is involved in implicit learning. This dissociation was made by studying Patient HM. As previously mentioned, Patient HM had a bilateral medial temporal lobe removal. This process removed most of his hippocampus, uncus, and amygdala. As a result of this process, HM suffered from severe anterograde amnesia; he could not form new memories following the surgery (Scoville and Milner, 1957).
This was shown through a series of tests. To show that HM’s impairment was memory-specific, Scoville and Milner conducted an IQ test. HM’s performance was standard and demonstrated that the impairment only affected his memory. Next, HM was asked to repeat a sequence of digits immediately after hearing them. His performance was also normal here, suggesting that his short-term/working memory was unaffected. To see if his long-term memory was affected, HM was given short stories. Then, he was asked to repeat them immediately and after a delay. While he was able to repeat them immediately, he showed very poor performance after a delay. This confirmed that his long-term memory was impaired. This was confirmed through word pairing tests. Furthermore, this impairment was not limited to verbal information; HM showed poor performance in spatial memory tasks as well. However, he did show vivid memories of his past, suggesting that his amnesia was anterograde rather than retrograde. These results suggest that the hippocampus is critical for the creation of new memories, and that long-term memory is separate from intelligence, reasoning, and working memory (Scoville and Milner, 1957).
Declarative versus Non-Declarative Memory
Scoville and Milner’s original study on HM was key to understanding the role of the hippocampus in long-term memory formation. However, a further study by Milner suggested that this observation was more nuanced than originally thought. In her 1961 paper, Milner asked HM to perform a simple task: trace a star by only looking at his hand through a mirror. The point of this test was to see if HM could learn a new skill. Surprisingly, HM’s performance improved over the days he practiced the task. This showed that memory for motor skills is part of a different neural circuit than memory for declarative information (Milner, 1962).
This was the first evidence that long-term memory could be split into multiple memory systems (Squire, 2010). To further investigate and formalize this theory, Cohen and Squire tasked amnesic patients with a mirror-reading task. Subjects were asked to read the mirror image of a set of words. The time it took them to read the words was tracked and recorded. As with HM, the patients were able to improve and learn the skill, despite having no recollection of having done the task. This led Cohen and Squire to formalize their theory of declarative versus procedural memory. While amnesia primarily affected declarative memory, it seemed to leave procedural memory alone. This showed that the medial temporal lobe is critical for forming declarative memories, whereas other areas of the brain are important for procedural memory (Cohen and Squire, 1980).
This theory was further refined by other studies on Patient HM. In one study, HM was given a set of words. Importantly, he was not asked to remember them. Next, he was given the first few letters of words and asked to complete them. While many words could be completed with those letters, HM was primed to write down the words he had seen. Despite not being able to remember the words themselves, he was still able to be primed with information (Gabrieli et al., 1994). This suggested that the split between declarative and procedural memory was better described as a split between explicit (declarative) and implicit (nondeclarative) memory.
Episodic versus Semantic Memory
At this point, it was evident that declarative memory formed its own subset within long-term memory. However, the question of further dissociation remained. This was tackled by Tulving in 1972, where he suggested that declarative memory could be split into two types: semantic and episodic (Tulving et al., 1972). He proposed this based on the results of experiments he performed, where he observed that memorizing context-dependent information (episodic) was different than memorizing facts which did not depend on context (semantic). For example, remembering what you ate for dinner two days ago would be considered episodic while remembering the types of declarative memory would be semantic.
Tulving’s theory was further proven through Patient KC. Like HM, KC had amnesia as a result of damage to the medial temporal lobe. However, as his impairment was due to a motorcycle accident, his hippocampus was never fully removed. This led to one major difference between KC and HM: KC had preserved semantic memory. KC was able to remember facts from his childhood and learn new ones. However, he could not remember any events or context-attached information. This confirmed the dissociation between episodic and semantic memory (Tulving et al., 1988).
The Dissociation of Nondeclarative Memory
As previously described, nondeclarative memory involves implicitly learned responses. HM was heavily involved in determining procedural memory and priming as components of this type of memory. However, there are other classes which also are encapsulated by nondeclarative memory. These include classical conditioning and habit formation. Several patients contributed to our understanding of these classes.
Classical conditioning consists of two types of responses: emotional and physical. Emotional memory enables us to learn about what we should fear from past experiences. As shown through Patient SM, however, this process can be impaired through damage to the amygdala. SM performs normally on IQ tests and unlike HM, does not show amnesia. However, she does not show responses to scary situations—live snakes, haunted houses, scary movies, among others. This lack of fear conditioning suggests that the amygdala is important for this aspect of memory (Feinstein et al., 2012).
The other type of classical conditioning evokes physical responses (eg. movement) due to a stimulus. One example of this is eyeblink conditioning. This is where a puff of air to the eye—which causes a blink—is paired with a neutral stimulus. After conditioning, the neutral stimulus alone is enough to cause blinking. However, this process is impaired within Alzheimer’s patients with damage to the cerebellum. This suggests that this area of the brain is important for conditioning of the skeletal musculature (Woodruff-Pak et al., 1990).
Another example of this is habit formation. This process involves repetition of a task, eventually leading to it requiring less cognitive load. As shown through a series of experiments, this process is highly dependent on the striatum (Knowlton and Patterson, 2018). Therefore, it is clear that nondeclarative memory is dependent on several areas of the brain outside the medial temporal lobe.
Conclusion
It is evident that encoding memories is a highly complex task for the brain. However, through studying impairments in patients, it is possible to piece its features together. This started off with Patient HM, who showed that the hippocampus is a critical structure for long term memory creation. Through Milner’s later study, this observation was refined to explicit memories. These explicit memories can be split into semantic and episodic memories. The dissociation of these two types was theorized by Tulving and proven through Patient KC. The other type of long-term memory, nondeclarative, was shown to consist of many types of learning: habit formation, classical conditioning, and procedural learning, among others. Several patients, such as SM, provided critical knowledge to our understanding of nondeclarative memory. By looking at this timeline, it is evident that our theories are constantly evolving and being further refined. Therefore, it is critical that we continue to study this field and make new discoveries about the mechanisms for memory formation.
Works Cited
Cohen, N., and L. Squire. “Preserved Learning and Retention of Pattern-Analyzing Skill in Amnesia: Dissociation of Knowing How and Knowing That.” Science, vol. 210, no. 4466, 10 Oct. 1980, pp. 207–210, https://doi.org/10.1126/science.7414331.
Feinstein, Justin S., et al. “The Human Amygdala and the Induction and Experience of Fear.” Current Biology, vol. 21, no. 1, Jan. 2011, pp. 34–38, www.ncbi.nlm.nih.gov/pmc/articles/PMC3030206/, https://doi.org/10.1016/j.cub.2010.11.042.
Gabrieli, John D.E, et al. “Dissociations among Structural-Perceptual, Lexical-Semantic, and Event-Fact Memory Systems in Alzheimer, Amnesic, and Normal Subjects.” Cortex, vol. 30, no. 1, 1 Mar. 1994, pp. 75–103, https://doi.org/10.1016/s0010-9452(13)80325-5. Accessed 18 Apr. 2025.
Knowlton, Barbara J, and Tara K Patterson. “Habit Formation and the Striatum.” Current Topics in Behavioral Neurosciences, 1 Jan. 2016, pp. 275–295, https://doi.org/10.1007/7854_2016_451.
Milner, B., 1965. Physiologie de l’Hippocampe: Colloque International, No. 107, Editions du Centre National de la Recherche Scientifique, Paris, 1962. 512 pp. 58NF.
Scoville, W. B., and B. Milner. “Loss of Recent Memory after Bilateral Hippocampal Lesions.” Journal of Neurology, Neurosurgery & Psychiatry, vol. 20, no. 1, 1 Feb. 1957, pp. 11–21, https://doi.org/10.1136/jnnp.20.1.11.
Squire, Larry R. “The Legacy of Patient H.M. For Neuroscience.” Neuron, vol. 61, no. 1, 15 Jan. 2009, pp. 6–9, www.ncbi.nlm.nih.gov/pmc/articles/PMC2649674/, https://doi.org/10.1016/j.neuron.2008.12.023.
Tulving, Endel, et al. “Priming of Semantic Autobiographical Knowledge: A Case Study of Retrograde Amnesia.” Brain and Cognition, vol. 8, no. 1, Aug. 1988, pp. 3–20, https://doi.org/10.1016/0278-2626(88)90035-8.
Tulving, Endel . “Episodic and Semantic Memory.” Psycnet.apa.org, 1972, psycnet.apa.org/record/1973-08477-007.
Woodruff-Pak, Diana S., et al. “Eyeblink Conditioning Discriminates Alzheimerʼs Patients from Non-Demented Aged.” NeuroReport, vol. 1, no. 1, Sept. 1990, pp. 45–48, https://doi.org/10.1097/00001756-199009000-00013.