If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains ***.kastatic.org** and ***.kasandbox.org** are unblocked.

Main content

Current time:0:00Total duration:5:03

Voiceover: For any certain
task, what we're going to have is a noise threshold. So, we're going to a have
a noise distribution, and the noise distribution might kind of look something like this. So, this is just background noise. So, what this basically
shows is if we were to take a bunch of individuals,
and we showed them a bunch, we experimentally tested
signal detection theory, you would get this graph,
just kind of indicating the noise, and then we'd get a
second graph, which is kind of shifted over to the right a little bit, and this is the signal distribution. So, this is the signal
distribution, and in the blue over here, we have the noise distribution. So, the difference
between the means of these two distributions is d-prime. So, if the signal distribution
was shifted over here to the right, then d-prime
would be really big. It would be a really easy task. It would be something kind of
like this, whether there is a green dot on the screen or not. But, on the other hand, if
the signal distribution were shifted over to the left,
then d-prime would be super small, and it'd be
something more like this, like the more difficult task. So, the x-axis, here we have
the intensity of the stimulus. So, that would be how
easy the stimulus is to distinguish from the background. Okay, so, we've got the first
variable, which is d-prime, and the second variable is C, and that is the strategy of the individual. So, strategy can actually be, strategy can be expressed via
the choice of the threshold. So, what does the individual
deem as necessary? What threshold is necessary to surpass, in order for them to say yes versus no? So, we're just going to label
the different strategies. So, there's B, there's D,
there's C, and there's beta. So, these are different
strategies, and they're just variables given to the
different strategies. So, if we were to use this B strategy, this would basically say, okay, I'm going to choose a certain threshold. So, let's say that I choose
this threshold over here. So, let's say that I choose two. So, anything that is greater
than two, I will say yes to, and anything that is less
than two, I will say no to. So, in that case, the probability of a hit is this area over here,
and the probability of a false alarm is this area over here. So, that would be the B strategy. The D strategy basically says, okay, so the position of my
threshold is going to be relative to the signal distribution. So, basically what that
comes out to be is d-prime. So, the signal distribution,
if it's over to the right, will have a big d-prime,
so d-prime minus B. So, we choose a threshold,
let's say two, and let's say that d-prime in this example is one, then we would have two
minus one, and we'd get one. So, if we were using the D
strategy, then anything above a one would get a yes,
anything below would get a no. So, the C strategy would
be an ideal observer. So, this would be someone
that would be minimizing the possibility of a miss
and of a false alarm. So, the C strategy is
basically, if we were to write an equation for it, it would
be B minus d-prime over two. So, in our example it
would be two minus one divided by two, so that is
equal to one point five. So, if you were using a C
strategy, that of an ideal observer, then we would say
anything above one point five would get a yes, anything
below would get a no. And so, we said over here
that the C variable is indicative of the strategy
that a person uses. So, when C equals zero, then the participant is an ideal observer. If C is less than one, we
say that the participant is liberal, and if C is greater than one, we say the participant is
conservative in their strategy. And this would be ideal. So, when we say conservative,
that means that they respond, no, more often
than an ideal observer, and when we say liberal,
then that means that the participant says no less
often than the ideal observer. So, the final variable that
we have to talk about is beta. So, if we're using this
beta virtue, we accept the value of the threshold
equal to the ratio of the height of the signal distribution to the height of the noise distribution. And so, it's easier to actually
look at beta in this way. So, if we were actually
writing an equation, we would say the natural log of beta is equal to d-prime times C. And so, in this case,
it would equal one times one point five, which
equals one point five.