Anterior Temporal Lobe

Anterior Temporal Regions: Amodal Storage of Semantic Information

To appreciate the role that anterior temporal regions play in memory, we need to consider that the ability to explicitly recall information or event is not limited just to specific episodes in our lives. Rather explicit (declarative) memories can be divided into two types: episodic memory Opens in new window and semantic memory Opens in new window (Tulving, 1972).

Semantic memory refers to knowledge that allows us to form and retain facts, concepts, and categories, both about the world and about the people we know, such as where they live, their occupations and interests, and their personality characteristics. Such information is not linked to a specific episode but rather pertains across many different episodes and context. In contrast, as discussed here Opens in new window, episodic memory refers to autobiographical memories about the time, place, and circumstances of a given specific experience.

To make this distinction clearer, consider an example of an episodic memory, the memory of your first kiss. This memory includes information about the person whom you kissed, the place where it occurred, how you felt, and so forth.

In contrast, your semantic memory about kisses includes information such as that they involve the placing of a person’s lips on someone else or an object; are used to demonstrate ardor, affection, or appreciation; and are commonly given when people are meeting one another or when they are leaving.

In this example, whereas information contained in semantic memory is about kisses in general, that contained in episodic memory is about a particular kiss. Episodic memory allows a reexperiencing of the event—providing the opportunity, in essence, to travel back in time (Tulving, 1985) whereas semantic knowledge allows us to generalize knowledge across time.

People with damage to the medial temporal lobe lose the ability to form new memories about episodes in their lives, suggesting a disruption of episodic memory. However, after injury they can learn at least some new semantic information, suggesting that semantic memories may not rely entirely on the medial temporal region (e.g., Bayley et al., 2008).

One of the most dramatic examples of such learning comes from three case studies of children who sustained damage in childhood (at birth, age 4, and age 9) that included portions of the hippocampus Opens in new window (Vargha-Khadem et al., 1997). Despite having a pronounced amnesia for the everyday episode that occurred in their lives, they were nonetheless able to attend mainstream school, at which they learned language skills (including the ability to read) and enough factual information to place their intelligence within the low average to average range.

Nonetheless, these cases are extreme ones in which the brain may have adapted developmentally to the lack of a hippocampus. We should not take too far the idea that new semantic information can be acquired totally independently of the hippocampus. In fact, as we have discussed here Opens in new window, after his surgery H.M., was not able to acquire new information about occurrences in the world, such as what an astronaut is, which clearly falls into the domain of semantic processing. Nonetheless, retrieval of semantic information may be somewhat possible without the hippocampal system.

How might that occur?

At least some aspects of semantic memory may rely on domain-specific neocortical processors. For example, your memory of the feel of wool is likely to be aided by reactivation of somatosensory regions, whereas your memory of the shape of sheep is likely aided by reactivation of visual areas (Binder & Desai, 2011).

But what about semantic information that is not linked to a particular modality?

Anterior temporal lobe regions may play a role in retaining such information (Simmons & Martin, 2009). Some evidence for this viewpoint comes from a disorder called semantic dementia, in which patients progressively lose the ability to retain semantic information. For example, when travelling through the countryside to visit a friend, a patient with this disorder reminded his wife where to turn to reach their friend’s house, but then looked at sheep in a field they were passing and asked her, “What are those things?” The most notable pathology in this disorder is degeneration of the anterior temporal regions.

Similarly, repetitive transcranial magnetic stimulation over anterior temporal regions in neurologically intact people slows the ability to name pictures (e.g., swan, poodle) and to pick synonyms for words (e.g., shown “rogue,” choosing “scoundrel” rather than “polka” or “gasket”). However, it does not interfere with nonsemantic judgments, like naming numbers or determining which numbers are closest in value (e.g., shown “36,” picking “40” as compared to “30,” “42” or “28”) (Pobric et al., 2007). Supporting evidence is provided by functional neuroimaging, which indicates activation in this region during tasks requiring semantic processing (Binney et al., 2010; Patterson et al., 2007).

Why, from a neuroanatomical position, would anterior temporal regions be well positioned to support amoda semantic processing?

Similar to the hippocampus, this region may serve as a convergence zone. But unlike the hippocampus, which is binding together information about specific episodes in time, the anterior temporal region is integrating sensory input from modality-specific regions with regards to specific episodes. For example, information from visual regions about the visual form of sheep and information from somatosensory regions about the feel of wool converge onto the anterior temporal lobe. As such, this region can over time codify and elaborate information about sheep in general such as that they look “puffy” because of their wool; that wool is warm; that because of its warmth, people use wool in clothing and so forth (Binder & Desai, 2011; Rice et al., 2015) (See Figure X).