Here is the "core concept of desire" proposed in chapter two of In Nature's Interests? On this conception, "A desires X" is true if and only if:

1. A is disposed to pursue X;

2. A pursues X in the way he, she or it does because A previously engaged or concurrently engages in practical reasoning about how to achieve X or objects like X, where engaging in practical reasoning includes both drawing inferences from beliefs of the form "Y is a means to X" and the hypothesis formation and testing by which such beliefs are acquired and revised; and

3. this practical reasoning is at least potentially conscious.

I propose that this captures the paradigmatic cases of desire, and I offer three reasons for working from it:

1. Something like it is accepted by many of the principals on either side in the animal rights debate (B 28).

2. Each of the three conditions helps explain why evolutionary biologists' attributions of desires to species are metaphorical (M 28).

3. In light of the evidence considered in the rest of the chapter, this definition explains why our intuitions about "real" desires get less strong and confident as we "descend" the common-sense phylogenetic "scale" (but see the qualifications about "phylogenetic scale" at pp. 31-32, and about octopi in particular at pp. 48-51).

Following are two tables summarizing findings about the distribution of some very basic learning capacities in the animal kingdom. These capacities seem relevant to possession of desires insofar as it seems implausible to attribute conscious planning to organisms which lack the capacities in question.

The summary is based on Martin Bitterman's pioneering research summarized in "The Evolution of Intelligence," Scientific American 212 (1965), pp. 92-100, coupled with a review of subsequent research.

For a full account of the cited evidence, see Gary Varner, In Nature's Interests? Interests, Animal Rights and Environmental Ethics (Oxford University Press, 1998), chapter two, "Localizing Desire."

Multiple reversal tests involve repeatedly reversing the reward pattern in simple learning experiments. For instance, a rat is first presented with two levers and rewarded for pressing the left lever instead of the right. When the rat has learned to press the left lever all the time, the reward pattern is reversed. Once the rat has learned the new reward pattern, it is reversed again, and so on.

• Progressive adjustment: This refers to the way rats handle this learning problem. They take longer to learn the new reward pattern the first time it is reversed, but on subsequent reversals, they learn the new pattern immediately.

• No progressive adjustment: Fish and "lower" animals do not exhibit progressive adjustment on these reversal problems.

Probability learning tests involve rewarding one of two alternatives a fixed percentage of the time, but randomly varying which alternative is rewarded on which individual trial.

• Matching: This refers to the strategy fish emply in such tests, which is to respond with the rewarded alternative exactly the percentage of the time that it is being rewarded, but randomly varying which answer they give.

• Maximizing: This refers to the strategy birds and most mammals employ, which is to respond 100% of the time to the more often rewarded alternative (which achieves a higher payoff percentage than matching).

• Systematic matching: This refers to a strategy which only mammals exhibit, namely choosing whatever alternative was rewarded on the immediately preceding trial (which is less likely to have a higher payoff percentage than the maximizing strategy, but which is a strategy which human test subjects explain by saying that they thought it might). Only primates ever exhibit the strategy of systematically responding with the opposite of what was rewarded on the last trial.