What is Natural Teleology?
Natural teleology is the ability to recognize that a fellow organism is a self propelled agent, and has goals. Not only does the natural teleologist assume that self propelled agents have goals, but they assume what those goals rationally correlated the agent’s observable behavior.
Natural teleology takes place before interpretation, both on the developmental timeline of a person’s life, and most likely in the evolutionary timeline of our species. Natural teleology is the precursor to true mental interpretation. You could even say that interpretation is just the most advanced and complex form of natural teleology. Like interpretation, natural teleology requires complex sensorimotor cognitive processes. The natural teleologist calculates a conspecific’s goals based on body language, facial expression, gaze, etc. Until interpretation came along, natural teleology represented what was the absolute frontier of social cognition in mammals.
What Isn’t Natural Teleology?
It is not true interpretation, in the theory-of-mind sense. The main distinction between the two is that the natural teleologist represents the goal states of the subject, while the interpreter represents the mental state. The line can get blurred at crucial stages of child development, due to the unpacking of new novel behaviors which make the child’s representations more mental. Natural teleologists do not have a theory of their own minds or others’ minds. They know their goals and they represent attributed goals to subjects who are in sensory contact with them. Out of sight, out of interpretation.
It is also not just recognition of crude agency. Crude agency recognition is present in all animals, and possibly all life forms. It is just the general disposition to treat other organisms as having goals. Crude agency can build into very complex procedural tasks for certain situations, but it does not have the multipurpose potential that natural teleology does.
Who Does Teleology?
The best example of different stages within the teleological paradigm is during the development of infants. Between birth and four years, every child follows a similar path from natural teleologist to interpreter. Each stage along the way contributes to the creation of a new form of natural teleology, which can be seen as a possible explanatory model for non human primates and proto-humans.
Non-human primates are our best glimpse into the world of lifelong teleologists. Whereas humans mature into interpreters, chimps, gorillas,bonobos and all the great apes seem to all get caught at different stages of the teleological path. Other animals might have natural teleology. Because of how evolution deals everyone cards that work in their environment, it’s also possible that in non-primates crude agency developed into a similar competency that isn’t necessarily natural teleology.
Innate Concepts- Baby Steps
Humans are natural teleologists for the first 2-3 years of life. There are several different stages and changes that take place, with true interpretation starting by age 4. It seems that the innate concepts that we have at just a few weeks of age constitute nearly all of the requirements of natural teleology. Infants at this age can rationally judge if the motion of an agent makes sense(Carey 2009). These tests show the most basic form of natural teleology employed by humans. Goal oriented, sensory contact needed, and very limited complexity of causal relations.
One Year Old- Giant Steps
At 12 months, we are adept natural teleologists, with some abilities which not even adult chimps have which emerge by 14 months. Many tests have shown how crucial this period is in child development, but also for understanding which cognitive abilities build off of which. By seeing which aspects of mental interpretation come first, we gain insight into how it works in an adult.
Gergely and Csibra(1995, 2003) define the abilities and competencies they are looking for, and try to draw several conclusions about the causal reasoning abilities of the infants. The tests that they compile clearly show that children as young as 1 year old can recognize goal directed behavior, and represent very simple novel propositions about what they see in front of them. None of this is highly groundbreaking, really. These tests are hardly more complex than the tests of naive physics carried out on children only a few weeks or months old (Carey 2009). This does establish the average 12 month old as a natural teleologist, but other tests have shown them surpass the minimum requirement of that title.
One of the more advanced abilities that infants have is imitation. Twelve month olds have been shown to be able to pick up behaviors with rational goals behind them with ease and consistency (Carpenter 2006.) By fourteen months they can do this with novel behaviors a week after first seeing the behavior (Meltzoff 1988). These demonstrate ways in which these infants have already passed apes.
These infants are also still lacking skills that can still be considered to be in the teleological skillset. They are highly limited in the complexity of causal relations they can represent and understand (Woodward et al 2000). This establishes them as still having more development to go within natural teleology before they approach interpreting.
At around 18 months, it seems that these infants have already begun to majorly advance towards the threshold of interpretation. This stage is still inherently natural teleology, but it seems to be more advanced than non-human primates in ways that show the importance of sophisticated interpretation on a complex social and cultural life.
1.5 year olds understand reciprocal altruism better than chimps (Warneken and Tomasello 2006). Experiments in which a “seeker” looked for a prize or food that 18 month old infants complied more often in helping the seeker, and chimps were especially less helpful when the seeker was trying to find food. This does not necessarily mean that these 18 month-olds were interpreting mental states, but it does mean that they grasped some social aspect of altruism better than the chimps. Infants after this age are the most advanced natural teleologists on the planet but are still not true Interpreters.
Between 2 and 3 years we see the emergence of more of these socially relevant understandings which start the process to representing mental states instead of just goals. At around 2, children start to pretend. Pretense is the first form of metarepresentation. But these children still do not pass the false belief test better than chance (Doherty) and they still cannot consistently talk about the pretend situation.
It makes sense to think of children being two-model teleologists. They still have their main single updating model but begin meta representation by just having one secondary model. These children have what some might call “prelief” (Doherty). This secondary model can serve as a pretend model of the world, and becomes the representation of the false belief of another person. This stage marks the end of natural teleology. Infants are now starting to be able to act as if (Leslie 1987). Acting as if allows them to attribute a false belief to someone else. Once they can entertain their second model and act as if this model was the primary model for another person, they have gotten to the basic idea of theory or simulationist theory of mind.
By four years of age, the teleological stance is largely gone. As they get better at attributing their second, or third model to what another person thinks, they are interpreters. They begin to perform much better on false belief tests, and understand metarepresentational materials and statements. They have not formed the same theory of mind as adults, but they have a theory of other minds. They Still do not have a theory of their own mind.
An infant’s development mirrors the evolutionary steps most recent to their species. This is specific to mammals, and applies most to humans. Since we are highly altricial, much of our mental development must take place during infancy. There is simply too much to know for us to be born with it. That’s why the best place to see real natural teleology in the animal kingdom is in the lives of the great apes. As our closest relative, they have presumably the closest form of natural teleology to us. Their abilities can shed insight into the stages that infants go through during development, and the stages our ancestors went through during evolution(Weiss and Santos 2006).
Chimps do not show even the rudiments of interpretation seen in human infants. They do not understand altruistic helping of non kin conspecifics (Warneken and Tomasello 2006) and they do not understand the highly important gesture of pointing as well as 2 year old babies(Tomasello 2006). These tests had a “helper” work with chimps and 2 year old humans. The helper would try several types of gesture, including gaze, to test which gestures helped the chimps and which ones helped the humans. The infants understood all of them better than the chimps. Chimps showed a very adept skill at tracking gaze, but could not perceive the pointing gesture very well. It seems obvious that the difference we see in these results is the understanding of joint or shared attention. This is a social skill that comes along with the tentative “two model” stage for infants. Chimps never acquire this ability it seems.
Despite this somewhat promising quote:
“Our results represent the clearest demonstration to date that chimpanzees know what individual groupmates do and do not ‘know’, that is, what individual groupmates have and have not seen in the immediate past. Other experimental approaches have either found negative results or found positive results only after extensive training”(Hare et al 2001).
these results suggest that chimpanzees may have a rudimentary interface for adding the gaze state, and goal state to get some kind of “what-they-witnessed” representation in relation to goal. They have very shaky understanding of how what a conspecific has witnessed will affect how that conspecific reaches a goal.
Primates are the only extensively tested and proven natural teleologists, but what if natural teleology is the specific result of the type of complexity commonly seen in primate social groups? What if another mammalian species developed a similar tracking mechanism that allowed for something like interpretation, but it wasn’t natural teleology. I can only hypothesize that maybe dolphins are more responsive to each other’s body language because there is less eye/face contact while swimming in groups than sitting in crowds like apes. Maybe this would still be natural teleology in that they are tracking goals, but dolphins would be more responsive to different types of goal behaviors.
The animal kingdom has several species which excel in intelligence, with humans being the leader of this elite group. With what we know about our primate cousins cognition, what can we theorize about other animals? Brain size and encephalization quotient seem to be good parameters for testing relative intelligence. These numbers show humans to have the highest predicted intelligence (Roth, Dicke 2005) but also show us who else is in the “club”. Elephants, whales, dolphins, monkeys, and walruses also show promising statistics. Maybe they aren’t quite at the level of chimps and infants, but chances are that these Are still natural teleologists.
The Spectrum Of Agency Recognition
Natural teleology is part of a long line of proto-interpretive abilities. All animals recognize crude agency, and I think this ability gets more advanced in two separate stages: Crude agency becomes procedural Crude agency, and Procedural Crude agency, presumably, becomes Natural teleology.
Procedural Crude Agency
Procedural crude agency is made up of highly complex conditional procedures. Observations of most mammals with herd or pack social structures show that these animals need complex, situated, real time procedures to react to certain agents( prey, predator, etc.). Why make this distinction? Just as natural teleology provides a precursor to interpretation, procedural crude agency is a precursor to natural teleology. This demonstrates where the “line” of natural teleology is. Once the procedural rules are so complex and situations are so varied that having a specific rule for every one is inefficient, Natural teleology is selected for, as a general purpose social cognition device.
The Larger Competence: Causal Understanding
There are obviously innate mechanisms which process causal interactions given certain assumptions. This is what natural teleology is born out of. But it also gives rise to the ability to manipulate hypothetical causal interactions, and conceptualize more abstractly.
Our non human primate cousins have an implemental view of causality. We can see this in their tool-making, in which they use an object for a certain purpose and never pay attention to its diverse causal properties. Just as their tool use is really just one use for one object, without change, they use each other like implements. The natural teleologist schemes and deceives a target with the same lack of causal understanding. In both cases the chimp is behind humanity because it is only utilizing a small amount of causal knowledge when it manipulates a tool or a conspecific. Humans more fully understand novel causal properties of conspecifics and of objects.
Natural teleology and interpretation are one in the same. There is nothing unnatural about true interpretation, and interpreters still represent goals. We are just so good at it that we have found a more useful descriptive term: mental states.
How can Natural Teleology impact artificial intelligence and machine learning? The most obvious insight is that just seeing which abilities infants acquire earlier and later, we can guess that the earlier competencies would be easier to instantiate in a program than the later one. If we do cross the major perceptual hurdles (programs for recognizing gaze, body language, facial expression etc.) would a highly intelligent computer be safer if it was not given the capability to translate between behaviors and mental states but only goals?
The major upside, I think, is that this type of program would never be able to start attributing mental states to itself. Then we don’t have a scary robot uprising scenario on our hands, but would it be as useful as a comparable computer running an interpretation program? Depends on what you want it to do I suppose. We don’t need robots that build cars starting to perceive mental states. In making robots that are capable of doing physical tasks and following orders, but do not need eerily human intelligence, the teleological model may be the most efficient and effective way to instantiate such a program.
Gergely, György, and Gergely Csibra. “Teleological Reasoning in Infancy: The Naı̈ve Theory of Rational Action.” Trends in Cognitive Sciences7.7 (2003): 287-92. Print
Meltzoff, Andrew N. “Infant Imitation After a One Week Delay.”Developmental Psychology 24.4 (1988): n. pag. Web.
Premack, David. “The Infant’s Theory of Self-Propelled Objects.” Cognition16th ser. 36.1 (1990): n. pag. Web.
Leslie, Alan M. “Pretense and Representation.” Psychological Review 94.4 (1987): n. pag. Web.
Gergely, Gyorgy, and Gergely Csibra. “Taking the Intentional Stance at 12 Months of Age.” (n.d.): n. pag. Web.
Schwier, Christiane. “Rational Imitation in 12 Month Old Infants.” Infancy 10.3 (2006): n. pag. Web.
Woodward, Amanda L. “Twelve-Month-Old Infants Interpret Action in Context.” Psychological Science 11.1 (2000): 73-77. JSTOR. Web. 04 May 2014. <http://www.jstor.org/stable/10.2307/40063499?ref=search-gateway:35c1c7e19ba30c39d7d8fe171a065506>.
Weiss, Daniel J., and Laurie R. Santos. “Why Primates? The Importance of Nonhuman Primates for Understanding Human Infancy.” Infancy(2006): n. pag. Web.
Warneken, F., and M. Tomasello. “Altruistic Helping in Human Infants and Young Chimpanzees.” Science 311.5765 (2006): 1301-303. Print.
Tomasello, Michael. “A New Look at Infant Pointing.” Child Development78.3 (2007): 705-22. JSTOR. Web. 05 May 2014. <http://www.jstor.org/stable/10.2307/4620661?ref=search-gateway:2817d6d863d6ad295bb8878964a8ad49>.
Tomasello, Michael. “Why Don’t Apes Point?” Roots of Human Sociality: Culture, Cognition, and Interaction (2006): n. pag. Web.
Roth, G., and U. Dicke. “Evolution of the Brain and Intelligence.” Trends in Cognitive Sciences 9.5 (2005): 250-57. Print.
Carey, Susan. The Origin of Concepts. Oxford: Oxford UP, 2009. Print.
Hare, Brian, Josep Call, and Michael Tomasello. “Do Chimpanzees Know What Conspecifics Know?” Animal Behaviour 61.1 (2001): 139-51. Print.
Andrews, Kristin, “Animal Cognition”, The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/win2012/entries/cognition-animal/>.
Samet, Jerry and Zaitchik, Deborah, “Innateness and Contemporary Theories of Cognition”, The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/win2012/entries/innateness-cognition/>.
Downes, Stephen M., “Evolutionary Psychology”, The Stanford Encyclopedia of Philosophy (Fall 2010 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2010/entries/evolutionary-psychology/>.
Ravenscroft, Ian, “Folk Psychology as a Theory”, The Stanford Encyclopedia of Philosophy (Fall 2010 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2010/entries/folkpsych-theory/>.