Wolf intelligence is a fascinating topic.  When wolves hunt, their behavior indicates that a great deal of forethought and problem solving ability is involved.  For example, they use social cooperation to conduct and perhaps “plan” coordinated attacks.  When running into a herd of prey, they continuously test the animals and make decisions on which single animal they eventually will kill.  They test them by smell, scattering the herd and looking for weaknesses, and sometimes by waiting.  After scattering a herd, wolves may stand and watch their prey, waiting for a weak individual to reveal itself.  Once a lone animal is selected, the wolves may chase it only so far.

During this pursuit, the wolves must decide to either continue their attack or call if off.  Their decision is based on many factors.  Does the animal turn and fight?  Is there an apparent weakness such as injury or illness?  They must also judge the degree of danger involved.  Is there risk of injury?  Is it worth the risk?  A hungry wolf may decide that it is.  A wolf’s life, however, is a hard one.  Food is not always available and perhaps great risks are taken often to obtain it.  Apparently, these are calculated risks.  Unnecessary confrontations are usually avoided.  An injury in the wild can be the same as death - it just takes longer.  Wolves, therefore, need to be good problem solvers and have the ability to anticipate many dangerous situations.

In other words, generalized rules can keep wolves out of trouble.  For example, not every hunt is exactly the same as any other, but they do have common factors (i.e., the prey’s behavior) that the wolves can use for predicting certain situations.  But what are generalized rules and why are they so important?  Virtually all mammalian species learn in this way, including humans.  Although there are few studies on wolf intelligence, dolphin research has provided us with a great deal of information regarding how mammals learn.

Pretend you are a dolphin.  Your trainer wants to know your capacity to form and use generalized rules.  He calls you to the side of the tank and lowers the following two symbols into the water:

You are asked to choose a symbol by pressing it with your rostrum.  You have no idea what this guy wants so you pick a picture and press the one to your left.  Your selection is wrong and you are not rewarded.  Your trainer removes the symbols and then lowers down two more:

You press the picture to your right. Wrong again!  "Who is this guy?  What is his problem?"  These symbols are removed and you are confronted with two new symbols:

You do not know why but you choose the symbol to your right. Correct!  You get a sardine and you are feeling happy.  You are presented with pairs of pictures for hours.  Sometimes you get a sardine and sometimes you do not.    After the 168th pair you realize that it is not the symbols' position or what it is doing that is important, but their shape.  You always have a choice between symbols that are circles and assorted other shapes.  You begin pressing the circles all the time and a sardine comes your way every trial.  It took hours to learn but the rule here is to press the circle for a reward.  You knew it all the time, right?

That night, you are resting at the water surface, full of sardines, and reviewing what you learned that day.  You are feeling pretty cocky knowing you struggled so hard and yet solved the problem presented to you.  In the morning, you swim proudly to the side of the tank and begin another day of testing.  Your trainer lowers the following pair of symbols into the water:

"What is this?  I am not going to float here for hours trying to figure-out what this guy wants.  Wait a minute.  I won't  need too!  I know from past experience that the shape is what is important.  I'll generalize what I learned yesterday to this situation.  The picture on the left is similar to a circle and I bet that is what he wants me to choose."  You press the oval and you are correct on the first trial.  "Wow, a mackerel.  My favorite!"

As a dolphin, you learned to distinguish between circles and squares on the first day.  It took a long time to understand because you did not know what the trainer wanted.  On the second day, you used your previous experience instead of trial and error behavior.   In effect, you learned to learn.  This is called a learning set.  The terms "learning sets" and "generalized rules" are synonymous.  You applied the concept of curvature or circles to a similar situation and with greater efficiency, solved a new problem.

In nature, few situations are similar in all respects.  Learning sets help explain how animals are capable of solving new problems without constantly using trial and error behavior.  Instead, animals use their past experience.  Most primates (including humans) and apparently dolphins are good at forming and using learning sets.  In fact, dolphins can learn to form and use rules in laboratory problems that present a variety of relationships (Herman 1986):

Stimulus - Reward:  Dolphin trainers use this relationship to train their animals.  The dolphins learn to perform behaviors on cue (stimulus), usually a hand gesture or sound.  When the behavior is performed properly, the dolphins receive a reward.  The general rule is complete the behavior the stimulus calls for and receive something nice.

Sameness - Difference:  Investigators test an animal's understanding of a concept with this relationship.  The test animal is asked to match an object (or sound) with the class of objects (or sounds) that it came from.  For example, if a dolphin was shown a small rubber ball then a bowling ball, it would answer "same."  If it was shown a fish instead of a bowling ball, it would answer "different."

Matching - Nonmatching:  In this relationship,  animals are presented with an object or sound.  This is removed and replaced with a pair of objects or sounds.  One of these is either identical or similar to the first object or sound and the animal must choose the identical one.
 

References
Herman, L. M.  1986.  Cognition and language competencies of bottlenosed dolphins.  In:  Dolphin Cognition And Behavior:  A Comparative Approach (Ed. by R. J. Schusterman, J. A. Thomas & F. G. Wood), pp. 221-252.  London:  Lawrence Erlbaum Associates.