Chapter 18: Why Rats Can't Go Grocery Shopping

Core Idea

The main rival to the social-brain hypothesis is the ecological-brain hypothesis: the fruit diet of early primates drove their brain growth. Ripe fruit is available only in short, scattered windows, so a fruit eater must plan before it is hungry. The Bischof-Kohler hypothesis said only humans can plan for future needs, but squirrel monkeys can and rats can’t. Bennett proposes that primates anticipate future needs with the same machinery they use for theory of mind. Imagining what your future self will want is like imagining what another person wants. Both require modeling a mind in a state you are not in now.

Frameworks Introduced

• The ecological-brain hypothesis:
  • Claim: frugivory (eating fruit), not social life, drove primate brain growth.
  • Evidence: many fruits are ripe and still on the tree for less than seventy-two hours. Some trees offer ripe fruit for less than three weeks a year. Easy fruits such as figs attract many competitors and vanish fast, while bananas in their hard skin have few. In 2017, Alex DeCasien at NYU studied over 140 primate species and found that being a frugivore explained relative brain size perhaps even better than social-group size.
  • Contrast: leaves, nectar, seeds, grass and wood last long and are not in sparse patches. Even carnivores rarely face short time windows for hunting.
• The Bischof-Kohler hypothesis (Doris Bischof-Kohler and Norbert Bischof, 1970s; named by Thomas Suddendorf):
  • Claim: other animals plan for current needs, but only humans plan for future needs.
  • Bennett’s revision: anticipating a future need is a harder kind of planning, and some animals that can plan (rats) can’t do it. But many primates can, not only humans.
• Anticipating future needs as theory of mind turned on yourself:
  • When to use: to explain how a brain can choose something it does not currently want.
  • How: the neocortex simulates options, and the older structures (basal ganglia, amygdala, hypothalamus) score them by how they feel now. Imagining pizza when full doesn’t excite the basal ganglia. To shop for Thursday’s movie night, you must model your future hungry self the way you model a hungry friend.
  • Why it works: the same ability to represent a mental state that differs from your own works for another person and for your future self.
  • Failure mode: your current state leaks into the model. Hungry shoppers buy more food for the week than well-fed shoppers, just as thirsty people think others are thirstier.

Key Concepts

• Frugivore: an animal that eats mainly fruit. Early primates picked fruit in treetops after it ripened and before it fell.
• Folivore: an animal that eats mainly leaves. Some primates are now mainly folivores.
• Planning for current needs versus future needs: getting to food when hungry versus getting food for next week when not hungry now.
• Hard-coded hoarding: mice store nuts before winter, but lab mice that never lived through a winter start hoarding when the temperature drops. This happens only in northern species, so it looks like an evolved response, not an understanding of future hunger.
• Primate forward planning: chimps choose nighttime nests on the way to popular fruits such as figs, but not to less contested fruits they like as much, and leave earlier for the contested ones. Baboons leave earlier when fruit is scarce.
• Evidence against “only humans”: there are anecdotal reports of chimps bringing straw outside to make a nest before they feel cold. Bonobos and orangutans choose tools for a task up to fourteen hours ahead. Chimps carry stones to places without suitable stones and make tools in one place for use in another.
• Mentalizing: building a generative model of one’s own mind (see the summary below).

Mental Models

• Use “Is imagining the mind of someone else really any different from imagining the mind of your future self?” as a test. Treat the answer as no.
• Think of your future self as another person whose wants you must infer (Figure 18.1). The figure shows a thinker with two thought bubbles. The left is labeled Theory of Mind and asks what James would want after not eating for 24 hours. The right is labeled Anticipating Future Needs and asks what I would want after not eating for 24 hours. Both bubbles show a hungry figure and a plate.
• Use planning before you are hungry as the defining demand of a fruit diet, unlike grazing or hunting when hungry.
• Think of the social and ecological hypotheses as two pressures acting on the same ability, not as rivals.

Anti-patterns

• Reading seasonal hoarding as foresight: mouse hoarding is triggered by cold, not by an understanding of future hunger.
• Assuming only humans anticipate future needs: squirrel monkeys, chimps, bonobos and orangutans show it.
• Shopping while hungry: your current state biases your forecast of your future state.
• Judging a future option by how it feels now: the rats’ logic, which Bennett sums up as “why give up extra raisins for water, I’m not even thirsty!”

Rules of Thumb (author’s wording)

• “When hungry, you overestimate your own future hunger.”
• “How can your neocortex want something that your amygdala and hypothalamus do not?”
• Suddendorf, quoted by Bennett: “a satiated animal may be unable to understand that it may later be hungry”
• “Our ability to make decisions today for faraway, abstract, and not-yet-existent goals was inherited from tree faring primates.”

Worked Example

Naqshbandi and Roberts’s thirst test (2006, University of Western Ontario).

1. Squirrel monkeys and rats chose between two cups: a small treat (a few dates for monkeys, a few raisins for rats) and a large treat.
2. Normally both species pick the large treat.
3. Dates and raisins cause strong thirst, often needing over twice as much water to rehydrate.
4. The rule changed. Choosing the large treat meant water only hours later. Choosing the small treat meant water 15 to 30 minutes later.
5. Squirrel monkeys learned to pick the small treat. They gave up food now for water they did not yet want.
6. Rats kept picking the large treat.
7. Conclusion: monkeys can act for a future need; rats can’t. Planning is not enough. You must also model a future mental state that differs from your current one.

Key Takeaways

1. A fruit diet requires planning around short ripeness windows and competition, before you are hungry.
2. Frugivory may explain primate brain size as well as or better than group size.
3. Squirrel monkeys anticipate future thirst and rats don’t, so the Bischof-Kohler hypothesis is too strict about “only humans”.
4. Anticipating future needs and theory of mind likely share one mechanism: modeling a mind in a different state from your own.
5. Both abilities appear in primates and not in many other mammals, and both show the same bias from your current state.

Connects To

• Ch 10–14 (Breakthrough #3, Simulating): the neocortex simulates options, and the older structures judge them by current valence.
• Ch 15: the social-brain hypothesis and the free time that fruit gave primates.
• Ch 16: projection bias in theory of mind (thirsty people think others are thirsty).
• Ch 17: the other repurposing of theory of mind, imitation learning.
• Ch 19–22 (Breakthrough #5, Speaking): what happened after the human and chimpanzee lines split.

Summary of Breakthrough #4: Mentalizing

Three abilities seem to have emerged in early primates:

• Theory of mind: inferring the intent and knowledge of others.
• Imitation learning: acquiring novel skills through observation.
• Anticipating future needs: acting now to satisfy a want you will have later but do not have now.

Bennett argues these may not be separate abilities. They may all come from one breakthrough: “the construction of a generative model of one’s own mind”, which he calls mentalizing. Figure 18.2 draws this as one source, mentalizing (modeling your own mind), with three arrows:

• Projecting your mind into the mind of another to infer their intent and knowledge gives theory of mind.
• Projecting your mind into the mind of another to simulate their motor skills gives imitation learning.
• Projecting your mind into your own future to anticipate future mind states gives anticipating future needs.

Evidence for one shared ability:

• The three abilities come from shared neural structures, such as the gPFC, that first evolved in early primates.
• Children acquire them at similar ages.
• Damage that impairs one tends to impair the others.
• Most important, they come from the same areas that let us reason about our own mind. These areas are needed for projecting yourself into imagined futures, for recognizing yourself in a mirror (damage causes mirror-sign syndrome), and for recognizing your own movements (damage causes alien-hand syndrome). A child’s ability to reason about her own mind tends to come before all three abilities.
• The Mountcastle constraint: besides size, the main change in the primate brain was new areas of neocortex. If every area of neocortex is made of identical microcircuits, as Mountcastle, Helmholtz, Hinton, Hawkins, Friston and others suggest, the new skills must come from a new application of neocortex, not a new kind of computation. A second-order generative model fits that constraint.

Two hypotheses, one coin. Dunbar argues the social-brain and ecological-brain hypotheses are two sides of the same coin. Mentalizing helped early primates both forage fruit and win at politics, so both pressures pushed the gPFC to grow.

Where the story stands. Our last shared ancestor with chimpanzees lived seven million years ago in eastern Africa. If the six hundred million years since the first brains were one calendar year, we are now at Christmas Eve, with seven “days” left. In that time our ancestors go from foraging fruit to flying Falcon 9 rockets.