Neuroanatomy of episodic and semantic memory in humans: A brief review of neuroimaging studies
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.105196
Source of Support: None, Conflict of Interest: None
One of the most basic functions in every individual and species is memory. Memory is the process by which information is saved as knowledge and retained for further use as needed. Learning is a neurobiological phenomenon by which we acquire certain information from the outside world and is a precursor to memory. Memory consists of the capacity to encode, store, consolidate, and retrieve information. Recently, memory has been defined as a network of connections whose function is primarily to facilitate the long-lasting persistence of learned environmental cues. In this review, we present a brief description of the current classifications of memory networks with a focus on episodic memory and its anatomical substrate. We also present a brief review of the anatomical basis of memory systems and the most commonly used neuroimaging methods to assess memory, illustrated with magnetic resonance imaging images depicting the hippocampus, temporal lobe, and hippocampal formation, which are the main brain structures participating in memory networks.
Keywords: Hippocampus, magnetic resonance imaging, memory, neuroanatomy, parahippocampal gyrus
Memory is an essential function for the survival of individuals and their species because the majority of cognitive abilities depend on the successful storage of information.  Memory has been defined as the process of encoding, storing, consolidating, and retrieving information. It is an emergent process in our brain, resulting from complex interactions between the biochemistry of neurons and their electrical activity in specific anatomical structures. 
Episodic memory (EM) stores information regarding events and their context, such as what the event was, and where and when it occurred.  It is important for clinicians to know the structures involved in this kind of memory because EM allows individuals remember who they are, their ideas, thoughts, and feelings. Without EM, we cannot learn from the past. It also gives us our identity and individuality, and is an elementary part of our consciousness. EM declines over the lifespan. The elderly take more time to store information, and their recall of recent events is less accurate than that of young adults. Alterations to the structures underpinning EM have been implicated in various disorders such as Alzheimer's disease,  epilepsy,  schizophrenia,  and depression. 
Studies in cognitive neuroscience have demonstrated that memory is a dynamic property of the brain as a whole, rather than being localized to any single region. ,,, New advances in magnetic resonance imaging (MRI) have allowed to obtain morphologic and functional evidence of the different brain structures involved in memory networks, which form interconnects across the entire brain.  Because memory depends on several brain systems working in concert across many levels of neural organization, modern neuroscientists can innovatively use MRI to determine the architecture of memory systems.
We present a brief review of the current concepts and classifications of EM and also show some selected gray-matter MR images from the hippocampus, thalamus, amygdala, temporal lobe, hippocampal formation, mammillary bodies, and cerebellum, as they participate in EM.
Memory is an important part of daily life, and allows us to effectively interact with the environment and other people. Memory allows for the retention of experiences, and it minimizes risks and facilitates the optimal use of the current environment. Memory consists of the encoding, storage, consolidation, and retrieval of information. ,,
Encoding refers to the initial registration or acquisition of information. It involves the capture of information by sensory systems and its conversion by neuronal coding for use beyond simple perception.  The nature of encoding may differ considerably depending on different memory demands.
Storage is the creation of a relatively stable memory trace or record of knowledge in the brain.  Such traces require neuronal networks that can engage in neuronal coding, which is the substrate of information storage, and is evoked when we remember specific information. ,,
Consolidation refers to slow physical processes, which continue after perception, that enable temporary changes in activity and synaptic strength to become long-lasting, and the later reactivation of neural activity to allow for the induction of long-term synaptic changes. This occurs both in the region where the representation was initially formed (e.g., the hippocampus) and in additional regions where the representation was initially weaker (e.g., neocortical structures), but which receive spreading neural activity from the hippocampus. 
Retrieval refers to accessing stored information.  It requires the reactivation of knowledge and is closely related to encoding. Any successful act of retrieval requires initial encoding and the persistence of information in the nervous system.  Because these processes require the active connection of several brain structures, memory is not dependent on a single anatomic structure to which we can assign responsibility for an isolated memory process. Rather, there are multiple memory systems that shape a widely distributed network of cortical and subcortical brain regions with specific neural behavior.
In the last few years, memory has been described as a network of connections whose function is primarily to facilitate the long-lasting persistence of learned environmental cues.  Memory systems are formed by organizing elementary structures, which consist of a neural substrate and its behavioral and cognitive correlates. Some of these structures are engaged in other memory systems, and they have emerged at different stages of evolution and at different stages in the development of organisms. 
Classifications of memory systems
Memory has been classified into many subtypes depending on its persistence, the contents of its stored material, and the presence or not of consciousness during learning and memory. 
In accordance with its persistence, memory was first separated into three sequential major systems, the sensory, short-term, and long-term memory systems.  Sensory memory allows for the recording of sensations and their storage in cortical structures.  Short-term memory was initially proposed as a precursor to long-term memory. However, it was later noted that not all information stored in short-term memory passes into long-term memory. In fact, one kind of short-term memory refers to information used exclusively for executing or developing other complex cognitive processes. ,,,,,,
Long-term memory has been divided by Tulving  based on whether the material is stored in EM or semantic memory. EM is defined as recollections of previous experiences from one's personal past, especially if focused on events that constitute our autobiographic memory. Semantic memory refers to general knowledge of facts and concepts regarding the world. This knowledge is not located in a specific time or place. 
Based on the contents of their operating characteristics, the kinds of information they process, and the purpose they serve, all memory systems have been classified as declarative and non-declarative [Figure 1]. Declarative memory refers to memories that can be consciously recalled, such as facts and events. It is divided into episodic and semantic memory. In contrast, non-declarative memory does not afford awareness of any memory content but does require consciousness, and it includes procedural memory, priming, simple classical conditioning, and non-associative learning. 
The quest to obtain an understanding of the functional organization of the normal human brain, using techniques to assess changes in brain circulation, has occupied neuroscientists for more than a century. William James, in 1890, studied changes in brain flow during mental activities. He based his work on the writings of Angelo Mosso, who wrote in 1881 that brain circulation changed selectively with neuron activity.  In 1950, Kety et al. provided the first glimpse of quantitative changes in blood flow in the brain related directly to brain function. 
Nowadays, cognitive neuroscience, a growing area in neuroscience, combines the experimental strategies of cognitive psychology with various neuroimaging techniques to examine how brain functions support mental activities.  The neuroimaging techniques used include positron emission tomography (PET) and magnetic resonance imaging (MRI). ,
These neuroimaging methods provide a means of measuring local changes in brain activity. PET uses radiolabeled tracers to visualize blood flow changes related to neural activity.  MRI can measure brain activity indirectly by taking advantage of a fortuitous physiologic property, which is that when a region of the brain increases activity, both blood flow and the oxygen content of the blood in that region increase.  This allows for the indirect visualization of neural activity through changes in oxygen content of the blood. MRI can also provide quantitative analyses of brain structure, so current quantitative MR reports may include measurements of segmented volumes from gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF), or even specific gyri.  Both functional and structural MRI have the advantages of being safe, providing excellent spatial resolution, and not requiring the application of any intravenous contrast agent. 
Anatomical basis of memory systems
In the last few decades, memory has come to be seen as networks of interconnected cortical neurons, formed by associations that contain our experiences in their connectional structures.  Likewise, we now believe that memory networks overlap and interact profusely with one another such that a cellular assembly can be engaged in many memory networks. These networks are integrated by specific anatomical structures such as the hippocampus, cerebellum, amygdala, frontal lobes, temporal lobes, entorhinal cortex, and basal ganglia  [Figure 2].
Brain structures associated with EM
Because EM deals with where information was acquired, this subtype of declarative memory depends on the integrity of the medial temporal lobe. The hippocampus has a critical role in EM because it contains place cells, head-direction cells, and grid cells, which are involved in the representation and recollection of spatial locations. The hippocampus is surrounded and connected with the entorhinal, parahippocampal, and perirhinal cortices, and also with several subcortical and cortical structures.
The left prefrontal cortex is relatively highly involved in encoding the information, whereas the right prefrontal cortex is more involved in the EM retrieval.  The hippocampal regions including the corpus of Ammon, dentate gyrus, and subicular complex, and the adjacent perirhinal, entorhinal, and parahippocampal cortices are essential for declarative memory  [Figure 3].
Brain structures associated with semantic memory
Semantic memory involves the semantic definition of objects. It uses shape, color, size, function, and motion information. A consensus has not yet been achieved regarding where this information is represented in the brain. Some authors have proposed that this information is stored in perceptual and motor systems, and was active when we first learned about an object.  In support of this argument, the occipital cortex is the beginning of semantic processing, which continues in the left temporal lobe [Figure 4]. The left inferior frontal cortex is relevant to word selection and retrieval.  The fusiform gyrus, located at the ventral surface of the temporal lobes, is highly active during naming and reading words.  In the Squire model, the fundamental memory structures are the hippocampus, neocortex, amygdala, cerebellum, and basal ganglia. 
The understanding of memory systems requires both isolated diagrams of brain structures and knowledge of the dynamic interconnections between these systems. Anatomical and physiological assessments of brain structures can be achieved with imaging techniques such as PET and MRI. Knowledge of the relationships between memory systems is not exclusively held by neuroscientists doing research in humans or animals. This knowledge is also available to clinicians currently involved in the study of brain pathologies affecting memory, such as neurologists, psychiatrists, geriatrists, neurosurgeons, and psychologists. Although it is still an evolving field, we believe that the evaluation of basic systems requires easy and straightforward methods. So, any clinician trained in these techniques could rapidly obtain reliable and quantitative information that is useful in clinical settings.
This study was supported in part by Medica Sur Clinic and Foundation.
Haydée G. García-Lázaro was scholarship recipient of the Institute of Science and Technology of Mexico City (ICyTDF number EJE11-86) and worked at the MRI unit of Medica Sur Clinic & Foundation from July 2011 to June 2012.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]