Normal healthy 

people have 


from time to time 

and it is not 


diagnostic of 

any pathology.



Memory, Cognition, Hallucinations
Cataloguing error-brain ideas
 By Jeanine DeNoma 

This article is based on talks given by Dr. Ray Hyman and Dr. Barry Beyerstein at the Skeptic Toolbox Workshop on “Human Error” held in Eugene, in August, 1995.

“I saw it with my own eyes!” is the credulous response to which skeptics often must respond. It can come as a dare: Are you going to believe my story or call me a liar? The correct response may be neither. “There is a third option,” maintained Barry Beyerstein. “Honest people can be mistaken.” Speaking at the CSICOP’s workshop on human error, Beyerstein, a neurophysiologist from Simon-Frasier University, and Ray Hyman, a research psychologist at the University of Oregon, outlined how our brains perceive the world and catalogued some of the errors that creep into our information processing.

     Our brains have three separate memory systems, explained Hyman. Incoming sensory information is stored for a matter of milliseconds in one storage area as nearly raw, unprocessed data. Our long-term memory acts like attic space, storing recently used items to the front and pushing old things to the back. Both of these systems feed into our short-term memory which acts as the theater for our immediate conscious experience.

     Short-term memory has a very limited capacity. Only about seven “chunks” of data can be held there at any one time, said Hyman. A data “chunk” might be, for example, seven random letters, words or objects. We partially overcome this limitation by grouping information and storing data in larger and larger chunks. This seven item limit, however, is a serious bottleneck, creating competition among incoming sensory information. Much of the redundant sensory data simply bypasses this bottleneck to reserve the limited available space for novel information. While sensory information is coming in, our short-term memory is simultaneously being fed information from our long-term memory. By the time information is actually coded into short-term memory it has received input from previously stored memories.

     “The old adage ‘seeing is believing’... can just as accurately be turned around to ‘believing is seeing,’” said Beyerstein. This no longer contains the assumption that what you see is necessarily out there. Standard cognitive theory says that what you see is affected by what you expect or want to see.

     This raises the question of whether we see, or imagine we see, things. From the view point of contemporary cognitive psychology, this is the wrong question to ask, said Hyman, because under the cognitive model everything we do is a combination of conceptually driven and sensory driven processes.

     Ordinary perception includes more memory than we normally give it credit for. Even very simple perceptions interweave sensory data, emotions and cognitive information. “Memory is more reconstructive than it is reproductive,” said Beyerstein. That construction is affected by attention, arousal, fatigue, boredom, expectancy, suggestion, prior knowledge, context and social factors. For example, studies have shown that hungry people are more likely to “see” food in ambiguous stimuli.

     Beyerstein posed the interesting, but still unanswered questions: Why do neural representations of objects in the brain feel like they are out in the “real world” rather than in our head? Why do perceived objects feel real? Why do things which are not there sometimes feel real? Why do things which are there sometimes feel unreal?

     “Our conscious awareness takes the richest, most stable, most complete, most vivid and coherent model available at any given time and ... that’s going to be our reality,” said Beyerstein. There must be a template of past information from which to compare, so things do not feel completely novel; and there must be a mechanism for choosing among competing models of reality.

     In most cases, the most stable model is the one built from our sensory inputs. However, under certain stressful, unexpected, frightening or traumatic situations we will lack the resources to maintain sensory input, Beyerstein explained. Input from memory may then become the most stable and coherent source, and it will fill the vacuum created by sensory loss in the form of imagery. This is likely what is occurring in “near-death experiences.” Extreme mental or physical trauma increases memory input and imagery, creating the hallucination.

     Because our imagery and perceptual systems simultaneously compete for the same hardware, differing mixes of internal and external data come into our consciousness. We can have differing degrees of perception, imagery and hallucinations occurring here. A hallucination is “any perception-like experience occurring in the absence of an objective stimulus,” said Beyerstein. By definition, a hallucination is detailed, complete, emotional and feels “real.” Hallucinations can occur in all sense modalities, for example, as in the phantom limb sensations of amputees.

     Despite the bad connotations associated with hallucinations, “normal healthy people experience these things from time to time and it is not necessarily diagnostic of any pathology or prediction of future pathology,” Beyerstein said. For example, we spend one quarter of our sleeping time in vivid hallucinations. We call them dreams. “And there is nothing I can find in the fine print that says dreams only occur in our sleep.”

     For hallucinations to occur, we must shift from language-based to picture-based information processing, there must be an increase in internal input and/or a decrease in sensory input, and our willingness or ability to check reality must be impaired. Anything that impairs our information processing and conscious control increases the chances for hallucinations, explained Beyerstein.

     Hyman described other types of errors that commonly occur. Because of the constraints imposed by our short-term memory, we must select what we are going to pay attention to. Errors occur when we do not deal with relevant information. All incoming information is ambiguous and under-represented in our short-term memory, so we must use context to determine what we think we hear or see. Contextual errors are common.

     Retrieving information from long-term memory can be a source of error. Also, the way we use memory can produce errors. We know, said Hyman, that children have poorer memories than adults, probably because adults make more connections with already stored information. Children generally rely on repetition. “Repetition is the worst way to get information into memory, elaboration is the best,” said Hyman.

     Coding our analog experiences into digital language forces us to simplify our experiences. This simplified view then feeds back into our remembrance of the experience. Language rehearsal requires us to go through an interpretive process and changes our memory. For example, said Hyman, the person who says, “I saw a light in the sky,” will remember seeing a light, while the person who says “I saw a UFO,” will remember a UFO.

     Environmental input errors occur when incoming information is corrupt: the old “garbage in, garbage out” problem. “As information accrues in any field, the amount of good information increases ... but bad information increases at a faster rate,” explained Hyman. The amount of misinformation from legitimate sources is increasing - even in science. Frequent exposures to any information increases its credibility and believability. This occurs even with information originally obtained from irresponsible media sources which, at the time, we dismiss. It occurs because the information’s content is stored separately from the information’s source: We forget where we heard the information. This phenomena is called “source amnesia.”

     “It is a total fallacy to think our brain is just a video recorder which takes in everything there is,” said Beyerstein. “It is a wonderful organ and has an incredible storage capacity, but it is not infinite ... it has to filter, take little snippets and store them. When you recall, you are not just playing back a video, you are playing back those stored signals ... and then inference fills in the rest.”

     “Our sense of reality is essentially a constructed illusion,” said Beyerstein.

Return to Archive Index
© 2000 Oregonians for Rationality