The Consolidation Debate
Written by: Scott D. Slotnick
The hippocampus binds information between different cortical regions during long-term memory. However, long-term memories may only depend on the hippocampus for a limited time. In the standard model of memory consolidation, a long-term memory representation changes from being based on hippocampal-cortical interactions to being based on cortical-cortical interactions, which takes a period of somewhere between 1 to 10 years (Alvarez & Squire, 1994). The estimated length of time it takes for consolidation to occur is based on patients with lesions to the hippocampus. For example, a person with hippocampal damage due to temporary lack of oxygen might have impaired long-term memories for approximately 1 year before the time of damage, which is called retrograde amnesia, and have intact long-term memories for earlier events. This suggests that the hippocampus is involved in long-term memory retrieval for approximately 1 year, as more remote long-term memories no longer depend on the hippocampus so they are not disrupted.
Although the standard model of consolidation has been very influential, there is a convincing body of evidence that indicates patients with hippocampal damage have disrupted long-term memories, particularly autobiographical memories, for events that occurred over 30 years before the time of hippocampal damage (Nadel & Moscovitch, 1997). This suggests that the hippocampus is involved in long-term memory retrieval for our entire lifetime. Proponents of this model argue that some measures used to assess long-term memory are not sufficiently sensitive (Nadel & Bohbot, 2001). To illustrate, a patient with hippocampal damage may have intact semantic/fact memory for public events from 10 years ago (e.g., they could answer the question, ‘which country did George W. Bush declare war on based on the conjecture that there were weapons of mass destruction?’), but may have impaired autobiographical memory (e.g., they could not remember visiting Disneyland 15 years ago). If only semantic memories were evaluated, the test would be less sensitive to measuring memory disruption, because semantic memories are less susceptible to disruption following damage to the hippocampus (Nadel & Moscovitch, 1997; Winocur & Moscovitch, 2011). In one fMRI study, participants answered questions about news events that had occurred in the past 30 years, which assessed semantic memory (Smith & Squire, 2009). The magnitude of hippocampal activity systematically decreased as a function of news events that were 3 to 6 to 9 years old and then mostly leveled off, which suggests the hippocampus is most active during memory for recent events. This evidence was taken to support the standard model of consolidation, where the hippocampus is only involved during retrieval of more recent memories. There are two problems with this interpretation. First, only semantic memory was tested. As the hippocampus appears to be primarily involved in detailed long-term memory retrieval such as episodic memory, the finding that the hippocampus is not as important for remote semantic memories is not surprising – the investigators used a test that was not sensitive. Second, the activity in the hippocampus did not drop to zero for older semantic memories, but was well above baseline even for events that were 30 years old. This indicates that the hippocampus was involved in memory retrieval for this entire period (and also suggests this region plays some role during the semantic memory task employed). If the hippocampus was no longer involved, the magnitude of activity in this region would have dropped to zero for remote memories. Thus, even evidence from proponents of the standard model of consolidation appears to support the alternative hypothesis that the hippocampus is involved during long-term memory throughout the lifetime.
Although much of the evidence appears to support the involvement of the hippocampus during long-term memory for events that occurred decades earlier, such evidence can be difficult to interpret because of uncertainty about the lesion location(s). Lesions are rarely restricted to the hippocampus. Traumatic events that damage the hippocampus, such as lack of oxygen, often damage other regions such as the dorsolateral prefrontal cortex. If damage did occur to multiple regions, it is uncertain which damaged region caused the disruption in memory performance. It could be the case that damage to the dorsolateral prefrontal cortex, which is known to be important during long-term memory retrieval, is the cause for deficits in long-term memory. This is why lesion evidence is most convincing when it is restricted to the one region of interest. One study reported results from patients with focal lesions restricted to the hippocampus following transient global amnesia (TGA), which is a temporary loss of long-term memory that is often triggered by a highly emotional or physically arousing event (Bartsch, Döhring, Rohr, Jansen, & Deuschl, 2011). TGA patients had disruption of autobiographical memories that were both recent (less than 12 months old) and remote (more than 30 years old).
There is a growing body of evidence that the hippocampus is involved in long-term memories throughout the lifetime. As such, the process of consolidation does not appear to result in the complete transfer from hippocampal-cortical memory representations to cortical-cortical memory representations. This is still a highly debated topic and will continue to be hashed out for years to come.
Check out the rest of the memory series blogs:
Part 3: The Brain Basis of Forgetting
Part 4: Episodic Memory in Mammals
Part 5: The Consolidation Debate