Experimental Research paper

It's a common intuition that memory is like a video camera: we store a copy of whatever we were experiencing, and later we remember by playing it back. Among the many reasons this view of memory is incorrect is that it assumes our memory is always, or at least very often, accurate. But in fact, not only is memory inaccurate, it's often less accurate than we think: we might be very confident in a memory only to realize it was totally wrong.

One way memory can be distorted is through misinformation. Loftus and Palmer (1974) showed participants a video clip of a car accident, then asked them to estimate the speed of one car in the video. However, one group of participants was asked how fast the car was going when it "smashed into" the other car, while the second group was asked how fast the car was going when it "hit" the other car. Participants who were asked the "smashed" question estimated much higher speeds than those asked the "hit" question, even though both groups of participants saw the same video--suggesting that memories can be distorted by how questions about the memory are framed. Worse still, we can be poor judges of our own memory accuracy: in one study of vivid, emotional flashbulb memories, confidence in the memories was extremely high, but those high-confidence memories were not any more accurate than typical memories (Talarico & Rubin, 2003).

We can also experience entirely false memories. In the Deese-Roediger-McDermott paradigm (Deese, 1959; Roediger & McDermott, 1995), participants see lists of words that are all related to a single "critical" word. For example, if the critical word was candy, participants might see words like sweet, chocolate, or bar. The important part of the paradigm is that the critical word ("candy") is never presented to participants. Despite this, when asked to recall words later, many participants will falsely remember seeing the critical word, and most will be confident that their recollection is accurate.

Mostly, these misinformation and false memory effects happen because remembering is reconstructive. We don't just press "play" on a video recording of the event; when we remember our brain is filling gaps and details and context every time we recall the event. Remembering is an active process, not the passive "playback" of recorded information, and that has practical implications. Consider that, when polled, people serving on a jury overwhelmingly indicate that eyewitness testimony is the most compelling evidence when trying to decide whether a defendant is guilty. But based on false memory and other memory distortions, shouldn't jury members be much less trusting of eyewitness testimony than they are? If confidence is not as associated with memory accuracy as we think, then it becomes difficult to determine whose testimony to believe. And moreover, if memory is reconstructive, then the mere act of recalling the event may distort the memory, further clouding the issue.

The basic trial sequence for the encoding phase is as follows. Instructions are presented at the beginning of the experiment.

The basic trial sequence for the Recognition test is as follows.

For each participant, three tab-delimited text data files are saved in the Logfiles folder. The .log file (filename "Subject-Experiment Name.log") is a standard Presentation logfile and contains detailed information about every event and response that occurred during the experiment. The summary file (filename: "Subject-Experiment Name-Summary.txt") contains simple summary statistics (e.g., accuracy, RT) for relevant experiment conditions. The remaining file contains trial-level data. This is the file that would typically be used for running simple analyses. A brief example of this file and description of the column headings follows. Note that two tables are printed, one for encoding trials and one for test trials.

Column heading list for Recognition trials:

Stimuli are taken from a tab-delimited text file. The text file should contain 16 columns. Each row represents one word list, with the first column being the "critical" word, and the subsequent columns being the related words, listed in descending order of relatedness.

The table below describes how port codes are assigned to responses and stimulus events (if port codes are sent). In general, responses will have port codes less than 10, and stimulus events will have port codes 10 or higher. Note that based on parameter settings, some of the events listed below may not occur in the experiment.

Translations are included for the following languages: English, Spanish, German, French, Chinese, and Japanese. Use the Language parameter to select an available translation. A 'Custom' language file (and associated stimulus files, if necessary) is included for all experiments that can be modified to create new translations. Some experiments contain captions that should be translated manually as part of a parameter (for example, a statement describing a target stimulus or position). For this experiment, you should check the following parameters for those captions: Correct Feedback Caption and Incorrect Feedback Caption .