Let us explore the common concept of memory: Most people are probably talking about 1. all the things I can consciously recall. and 2. think things are forever stored in their mind like a filing cabinet. The above 2 points are very general, and I admit they may be wrong.
Lets look at memory moving from behaviorist to more cognitive understanding: Classical Conditioning:
Pavlov makes a dog salivate when ringing a bell. Memory or recall at an unconscious level is responsible for the reaction.
Operant Conditioning: Punishment for a particular behavior will reduce the likelihood of that behavior occurring again. Again memory is somewhere in the mix. I have a response I will do less often because I don't want to get punished.
From the above orientation, we see memory as being inherent in the behavior.
Moving more cognitively: People who have damage to their hippo campus can develop anterograde amnesia. These people can not form new memories. (think Memento). Studies have been done on them to determine if indeed they no longer have the ability to produce memories. One experiment, used an aversive event on subjects, then distracted them long enough that they forgot the event. It was found that even though they would forget the event, they would still physiologically react to it before it occured again. In this case the reaction would be increased heart rate and sweat.
More cognitively (and more salient to this topic): Elizabeth Loftus has found that memories can be changed after the fact. She has found that that certain ques in questions can change the way we remember things. For instance, in one experiment participants viewed a car accident on video and then answered a question about how much broken glass was on the street at the accident. Participants who were prompted with the question "How much glass was on the street" reported remembering glass significantly more often than those who were ask "Was there glass on the street?" This shows that memory retrevial is fallible.