Let's summarize what's
Signal Detection and Decision Theory
- When positive and negative signals overlap, mistakes in responding
to the signals are inevitable. For example, you cannot decrease the
number of "missed detections" without increasing the "false
- The utility of responding to any particular signal can be modeled
by a function incorporating three types of signal parameters: the distribution
of the signals, the relative abundance
of the signals, and the costs/benefits of the four payoffs.
- These signal parameters must be taken into account by an animal when
deciding how to respond to the signals.
- Changing a signal parameter changes the utility of responding to a
Stimuli Discrimination Learning
- When trained to respond to a stimulus, and then presented with a range
of stimuli including the training stimulus, many animals will "generalize"
and respond to some of the new stimuli that are close to the training
- Animals trained to respond to a stimulus and to not
respond to a nearby stimulus, when presented with a range of stimuli,
will shift their peak response away
from the negative stimulus, so that it is not centered on the positive
A Signal Detection Theory Interpretation of Peak
- Stimuli can be thought of as signals.
- The "peak shift" is predicted by the utility function based
on the parameters of signal detection.
- Peak shifts away from the negative signal occur when the cost of making
mistakes is increased, either by manipulating the payoffs, or by increasing
the overlap of the two signals.
- Like many other animals, bumblebees will generalize from a trained
signal to a range of nearby signals.
- The peak shift phenomenon has almost never been studied in invertebrates
(such as insects).
- Bumblebees displayed the predicted peak shift when the relative abundance
of the positive signal was lowered, and when the payoff for correct
detection was lowered. Both changes increase the cost of mistakes.
End of Bumblebee Signal Detection module.