![]() ![]() Similarly, a False Alarm rate of 16% means that 16% of the “Left” decision variable distribution must be above the criterion, or 1 standard deviation above the mean of the “Left” decision variable distribution. For example, a hit rate of 84% means that 84% of the “Right” decision variable distribution must be above the criterion, or in other words, the criterion is located 1 standard deviation below the mean of the “Right” decision variable distribution. Similarly, we know how far the “Left” decision variable distribution is shifted downward or leftward relative to the criterion by the percent of False Alarms*. Given enough trials, we know how far the “Right” decision variable distribution is shifted upward or rightward relative to the criterion by the percent of Hits. Because we’re assuming that underlying decision variable distributions are normally distributed, we can make inferences about how far apart the different distributions are by looking at the patterns of Hits and False Alarms. If we arbitrarily define rightward angled patches as our “signal” distribution, the different types of responses can be seen below. Incorrect guesses that a signal is present when it really isn’t are called “False Alarms”, and guessing a signal was absent when it really wasn’t are called “Misses”. Similarly, when they correctly say that no signal was present, this is called a “Correct Rejection”. When a person correctly detects a signal (flash of light or dim tone), this is called a “Hit”. In classic signal detection theory experiments, scientists were studying the limits of human perception and so were measuring whether people could see a dim flash of light, or hear a faint tone. ![]() But some of the time we’d choose the incorrect rotation by chance alone. The idea is that when the decision variable is above a certain criterion, we should respond “right”, and if it’s below a certain criterion, we should respond “left”. The next core idea of signal detection theory is the criterion. Similarly, if a Gabor patch is rotated slightly to the right, the corresponding decision variable distribution might look something like this: The decision variable distribution when a vertical Gabor patch is shown should look something like this: The shapes shown above are called Gabor patches, and they’re often used in studies of visual perception because they’re precisely defined, and different aspects of the shape can be changed independently (e.g. Further, it’s assumed that the variance in decision variables from one presentation of the stimulus to the next will be normally distributed (i.e. Regardless of what brain activity might correspond to a decision variable, it’s assumed that given the exact same stimulus, there’s going to be some variation in the corresponding decision variable. There’s been a great deal of research studying what the “neural correlates of a decision variable” are for different types of decisions. When we make judgments about physical stimuli, we’re actually making judgments about the neural activity associated with those stimuli. Simply put, when stimuli change, our brains process the stimuli differently. The premise here is that as certain features of the physical stimulus change, there is a corresponding change in the neural representation of that stimulus. The first core concept in signal detection theory is the idea of a decision variable. After all, isn’t preference just a type of perception, as in, perceived benefit, or perceived value? Most importantly, the same scientific rigors used to measure perceptions can be used to measure preferences. Whether we’re trying to detect whether a product change influences user satisfaction, or trying to measure preference for a product design, having a good understanding of signal detection theory serves as a foundation for future research method discussions. The ideas behind signal detection theory occur frequently in user research, so they’re worth reviewing here. The field of studying how physical differences in stimuli affect perception is called psychophysics, and one of the research techniques for studying this is signal detection theory. The process by which our nervous system converts physical stimuli to neural activity is referred to as sensation, but making sense of that information is referred to as perception. Is the shape below rotated to the right or the left? ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |