Reflexes that have themselves been conditioned in response to a conditioned response. For example:

First: We train a dog to salivate (the conditioned response, or CR) in response to a bell (the conditioned stimulus, or CS), based on associating the bell with the site of food (the unconditioned stimulus, or US),

Second: we train the dog to associate the salivation upon hearing the bell, with some new unconditioned stimulus, say, a circle being displayed in the dog's field of vision. In this case, the first conditioned stimulus (the bell), once conditioned, becomes the "unconditioned"** stimulus which is used to produce a new response to a new CS (a circle).

The Second association is the secondary conditioned reflex (salivating in response to a circle instead of the bell). For association to the alimentary reflex in dogs at least, Pavlov noted that the second association is about as far as you can push it. That is, you could not use the appearance of the circle to produce yet another salivation response based on some other conditioned stimulus. Pavlov sometimes referred to these as second-order and third-order reflexes.

Pavlov was able to produce third-order reflexes in defensive responses to noxious unconditioned stimulus, such as electric shock. But third order reflexes were as far as it went (with dogs).


. . . . . . .
Also Note

** When referring to the bell above, the word "unconditioned" is quoted because the bell isn't really an unconditioned stimulus. It merely serves as one, once that association has itself been taught. Pavlov established that even the site of food is actually conditioned, observing that it is an unknown stimulus in newly weened puppies, until the first time they actually begin to eat it. Therefor it can be thought of as a form of unconditioned stimulus in the strictest sense.