Evolutionary Reasons for Memory

Fallibility and the Evolutionary Reasons for Memory

A characteristic of long-term memory systems that is important to many fields, such as law enforcement and standard testing, is their relative stability or instability. Human memory is a sap that isn't always very sickly.

An example of this instability can be found in your answer to the following question: What were you doing when John Kennedy was assassinated? If you were around during that time, you probably believe that you can answer that question very accurately.

But how well is that long-term memory trace laid down during your experience of Kennedy's assassination? How about those of your friends? There's a way to test this. You could ask people to write down their activities just after some natural disaster occurred. You would then hold onto their answers for a period of time – years even – and then ask them to recall what they were doing when the disaster occurred. Your task would be to compare their years-later recollections to the responses made immediately after the dramatic event occurred.

Such research has been done. Memories of what people were doing when the space shuttle “Challenger” blew up, and during other events, have been assessed. Two consistent findings have emerged from such ghoulish research: People are very confident about the accuracy of their memories, and they are mostly wrong.

Memory degradation occurs very quickly with time, even for important memories. The 9/11 study showed that only two years after the tragedy, 55 percent of those queried were making errors in their autobiographical recollections. And it got worse with time. This was a surprise to nearly everybody who enrolled in the study. They thought that they had perfect recall of what they were doing.

The old memory no longer exists and is no longer accessible. Photo by Elena.

It's a sad fact for mots of us, but one of the first characteristics we discover about our memory traces is their susceptibility to erosion. Different types of memories efface at different rates. You easily forget 98 percent of the semantic memory content in a classroom 30 days after you've had the lecture.

Exactly how does that instability occur? Why does it occur? It all has to do with the mechanisms the brain uses to retrieve information. The interesting fact is that you have different retrieval gadgets, and you use them differentially over time.

Consider the case of semantic memory, or memory for things. When a semantic memory is first established, you are busy taking in the moment, and you can see all of the details. This doesn't require much effort. Your recall is robust.

This more stable form of recollection is sometimes called reproductive retrieval memory, because you are reproducing what you experienced on the spot. It doesn't stay stable, unfortunately. As time goes by, original pieces of the memory begin to fade. You lose small details of the memory trace and large details of the memory trace. You begin confusing information from other memory traces with the memory trace in question. The memory is eroding and getting confused.

If you want to retrieve this older memory, you are going to have do some reassembly. The old memory no longer exists or is no longer accessible. You will begin reassembling your memory from component parts as you recall strategies. We call this type of recall reconstructive retrieval memory.

Reconstructive retrieval is almost never as accurate as reproductive retrieval. Original details may get left out entirely as you attempt your reassembly. Details not in the old memory – often from other memories – may get mistakenly inserted into your reconstruction efforts. Astonishingly, you often don't recognize the error and swear they are part of the original.

As time passes, we move from reproductive memory to reconstructive memory. That's why the test subjects' 9/11 recollections failed two years out. At first, their recall was very accurate. But as time passed, the less-accurate reconstructive retrieval systems kicked in, complete with their 55 percent inaccuracy rates.

The most surprising thing about this research is out blind confidence that such a thing was not happening. But it does happen, made worse as we get older – at least for most of us. Oddly enough, the loss is uneven. Ts there something in our evolutionary history that favored retrieval systems to be unevenly fragile and increasingly inaccurate, always with limited initial capacity?

It's an inefficient design if memory is supposed to be providing us with an archive so that we can accurately remember the past. It erodes far too quickly to be a reliable library. Some researchers think that what's wrong with this explanation is not our memories but our conception of what purpose memories actually serve. Some researchers believe that we have misinterpreted the evolutionary intention of memory.

They think that our memory systems are not built to remember what happened previously so that we can remember yesterday. Instead, they are built so that we can make more accurate predictions about what might occur tomorrow. You don't need it to be photographic if you are trying to reimagine what might happen to you. Accuracy is only modestly required; you just need an approximation In this view, memory systems are gadgets for the future, not the past.

Memories vanish as time goes by. Photo by Elena.