The Hippocampus

The Role of the Hippocampus in Retrieval

Neuroimaging evidence indicates that activity in the hippocampal system Opens in new window occurs across a wide variety of retrieval conditions, including those requiring recall of episodic, semantic, and autobiographical information (Burianova, McIntosh, and Grady, 2010; Ryan et al., 2008). However, the portion of the hippocampus involved in retrieval may be somewhat distinct from that involved in encoding (Eldridge et al., 2005).

Meta-analyses across PET and fMRI studies suggest that more anterior regions of the hippocampus are engaged for encoding of episodic memories Opens in new window, while more posterior regions are involved in retrieval (see Lepage et al., 1998; Spaniol et al., 2009).

The exact nature of the processes or operations being performed by the hippocampus at the time of retrieval is still somewhat of a mystery. Nonetheless, there is evidence that the hippocampus participates in the reactivation of long-term memories. When some aspect or component of an event is experienced later in time, the hippocampal system allows for retrieval by linking that component with other component pieces, each of which is stored in distinct neocortical processors (e.g., Staresina et al., 2013).

In this manner, the hippocampus can be seen as allowing for pattern completion—one smaller piece of information can be used to reconstitute the whole (see Halgren, 1984; Eichenbaum, 2000).

You may have an intuitive sense of how this pattern completion works from your own experiences. For example, you may be heading out the door for the day and find that you don’t have your keys. By thinking back to the point in time at which you last walked into your house or apartment, a relatively small piece of information, you may be able to reconstitute the entire episode including where you placed your keys.

Evidence for such pattern completion comes from neuroimaging studies in which multi-voxel pattern analysis is used to ascertain the pattern of activity over the hippocampus during encoding Opens in new window. In one such study, people learned to classify specific nature scenes as falling into one of two random categories (A versus B), and the multi-voxel pattern analysis in the hippocampus associated with each category was determined (Bonnici et al., 2012).

People were then shown 50/50 morphs between the two scenes and had to judge whether the morph fell into category A or category B. If the decision is being made totally on the basis of perceptual information, the multi-voxel pattern should look like a blend of the two patterns. But if the hippocampus is doing pattern completion, then the multi-voxel pattern should mimic one of the two categories.

The researchers observed that if the person judged the 50/50 morph as belonging to pattern A, activity across the hippocampus was more similar to the multi-voxel pattern analysis associated with Category A than Category B. Conversely, if they judged the 50/50 morph to belong to pattern B, the pattern of activity was more similar to the multi-voxel pattern analysis associated with Category B than Category A.

Here we can see that the hippocampus takes some perceptual features from the 50/50 morph and uses them to complete the rest of the information for a given category. Some evidence suggests that the hippocampus’ involvement in memory retrieval may vary depending on the manner in which information is retrieved.

Some researchers have argued that recall of an item, as well as the recognition that an item has been seen before, rely on the same fundamental processes, but that the confidence one has or the strength of information is greater for recall than recognition (e.g., Dunn, 2004).

However, other psychologists have argued that two separate and distinct processes occur in memory retrieval. These models known as dual-process models, argue that recognition relies on the strength of undifferentiated information about the item or event, or, said more simply, a sense of familiarity. In contrast, recall involves remembering something specific about the item, such as the instance or episode in which the information was first learned.

In memory studies, experimental psychologists often ask people to differentiate the type of memory they have for an item by asking at test whether the person is familiar with and “knows” that s/he has seen the item before, as opposed to recalling the specific item and “remembering” that s/he has seen it before. Electrophysiological studies supply evidence for distinct neural signatures of each of these processes, suggesting that they are indeed separable.

In a typical experiment, people see a list of items to remember. Afterward, they are given a list of items and must decide whether they have seen the word before or not. Included in the list are new words, known as lures. Some of the lures are very dissimilar to words on the initial list. Others, however, are quite similar (e.g., “trucks” as compared to “truck”). Not surprisingly, people often report having seen these similar lures before, resulting in a false alarm.

One component, recorded maximally over parietal regions of the brain at about 600 ms post-presentation as a positive deflection, is greater to a correctly recognized item than to either similar lures that individuals claim they have seen before or dissimilar lures that they correctly identify as new.

Notice that this component distinguishes between items indeed seen before from those not seen before, regardless of what an individual reports s/he has seen!

This old/new parietal component may be indexing processes that are required when a person really remembers an item, such as accessing information about the spatial and temporal context in which the item was learned. In contrast, a negative component is recorded over frontal leads at about 400 ms that is larger for correctly rejected new words than for items that an individual thinks s/he has seen before, regardless of whether that is the case (e.g., studied items correctly identified as old and similar lures incorrectly identified as old). This component may reflect the familiarity component of memory retrieval.

Converging evidence from neuroimaging studies and patients with circumscribed brain lesions also suggests that different neural systems are engaged by familiarity and recollection (Skinner & Fernandes, 2007).

In addition, it is suggested that each of these processes relies on distinct neural substrates. Results from patients with damage to medial temporal lobe structures and neuroimaging studies suggest that the hippocampus and related midline diencephalic structures (e.g., mamillary bodies, anterior thalamic nuclei) are required for specifically remembering an item or event along with the complexity of its larger spatial and temporal context. Retrieval processes associated with familiarity, in contrast, are proposed to rely on the nearby perirhinal cortex and connections with the dorsal medial nucleus, which are likely to have representations that are not as precisely pattern-separated as those of the hippocampus.

An intriguing recent hypothesis suggests that the distinct nature of representations underlying recognition, compared to familiarity, can influence how we forget (Sadeh et al., 2014, 2016). Because hippocampal representations during encoding are designed for maximal pattern separation, they may be relatively robust to interference from the learning of new items.

The hippocampus does a good job of separating and distinguishing between items and closely associated lures. However, physiological research suggests that even in adults there can be neurogenesis of granule cells in the hippocampus, meaning that hippocampal circuits undergo remodeling. This remodeling may make it difficult to then reconstitute a memory via pattern completion (since the underlying neural circuitry required has been altered). As a result, with time, recognition memory will decay.

In contrast, representations underlying familiarity are more overlapping and less distinct, as exemplified by the research with lures that we discussed above. While such a mechanism allows new representations to be integrated with similar previous ones, the downside is that representations associated with familiarity are subject to interference from similar experiences or items, and forgetting will be driven by how many overlapping and confusable experiences occur.

Interesting, recent evidence suggests that because the hippocampus helps us retrieve information from the past, it is also involved in imagining the future. For example, neuroimaging studies show even more activation in the hippocampus when imagining the future than when retrieval of memories is involved (Addis et al., 2007).

Furthermore, patients with unilateral temporal lobectomy or hippocampal damage report fewer details when imagining potential future events (Kurczek et al., 2015; Lechowicz et al., 2016), and neuroimaging studies show increased hippocampal activity for the initial imagining of a future event, compared to the recall of a previously constructed imagined event (Gaesser et al., 2013).

What are the implications of these findings?

At least some researchers have suggested that a better ability to remember the past may be associated with better abilities to plan for the future. For example, larger hippocampal volume is associated with a better ability to think about and implement plans for the future (Jung et al., 2016).

Why might the hippocampus play a role?

The hippocampus might aid thinking about the future in two ways.

• First, the richer the store of information one can access from the past, the larger the repertoire of potential responses that one can call upon to imagine how one might act ini the future.

• Second, as we have learned, the hippocampus serves to bind together disparate pieces of information into a coherent whole. It also may be able to recombine pieces of information in a novel way to problem solve and aid thought about future action (Schacter & Addis, 2009).

As we mentioned earlier, Tulving (1985) remarked that memory, and, in this case, the pivotal role of the hippocampus, allows for time travel. The hippocampus appears to not only aid us in traveling back in time, but also traveling into the future.