|Reece Roberts, Aleea Devitt, Donna Rose Addis|
We’re a group of researchers in the Memory Lab at The University of Auckland. Our research interests are broad, ranging from autobiographical memory to false memory to imagination and creativity, and we approach these topics from both cognitive and neural perspectives. But the strand that threads all this research together is the constructive nature of memory.
It is well established that episodic memory is a constructive process. Seminal work from Bartlett (1932) demonstrated the fallibility of memory, and also highlighted the ways in which it can be distorted. Building on this earlier work on false memory, Schacter and colleagues proposed a framework for the cognitive neuroscience of constructive memory in the early 1990s that has proved influential. Since that time, advances in neuroimaging have provided evidence to support three key ideas: (1) that features or elemental parts comprising an episode must be linked together, probably by the hippocampus, to form a memory trace that we encode; (2) that these elements are stored in a distributed fashion across the brain in the regions that originally processed the information (e.g., face information in fusiform gyrus); and (3) that these elements must be reactivated (and re-integrated) upon retrieval.
Because memories are stored as elemental features that have to be pieced back together, there are many opportunities for this process to go awry. First, we have to try and keep different episodes with similar elements separate – a process called “pattern separation”; recent research suggests that the growth of neurons in the hippocampus (neurogenesis) might facilitate separation by providing new “clean” neurons ready for encoding a distinct experience. Second, when we retrieve a memory, the cues available to us (either in the environment or internally generated) are usually only partial. That is, we have to use the available cue as a trigger to find the rest of the elements and complete the memory – a process known as pattern completion.
While often times we successfully encode and retrieve a memory, these processes of separation and completion can – and do - fail to varying degrees. We might, for instance, not reactivate some of the elements (forgetting that John was Matt’s wedding), or not activate any of them at all! We could also incorporate a wrong element into a memory (‘remembering’ that Annie was at Matt’s wedding) or mesh together multiple elements from various distinct memories into an entirely fabricated memory.
While seemingly negative, the constructive nature of memory may actually confer some adaptive advantages. In particular, it may provide the foundation allowing us to imagine. But how do we use memory to imagine? How do we separate what is real from what is imaginary, and where does the boundary between the externally experienced and internally generated events lie? What factors influence whether we successfully encode and/or retrieve a veridical memory or a distorted one? And are any memories actually veridical? These are just some of the ideas we hope to explore in this forum.