Conceptual foundations of generalizations

    Generic statements are common ways of expressing everyday generalizations. Most people agree with generalizations such as, “dogs have four legs”, “sharks bite swimmers”, “shoes have laces”, and “ducks lay eggs”, all of which are generics. Those generalizations are based on an underlying network of concepts, and a new set of studies investigates the structure of that network.

    Artwork credits: Burak Arikan‘s 2012 piece, “Network Map of Artists and Political Inclination


    Generics are ubiquitous in daily conversation, and their use is no more evident than in an election season, when pundits are likely to describe the properties of entire voting blocs. For example, you might hear the following generalizations echoed by members of the newsmedia:

    • Women are against cuts in Planned Parenthood funding
    • Floridians are concerned with immigration policies
    • Republicans and Democrats both want to overhaul the tax system

    Each of these statements is a generic assertion, i.e., a statement of the form “Xs are Ys.” Despite their ubiquity in daily speech, generic statements pose a challenge for linguistics and philosophers because they are hard to classify in a uniform way (e.g., Krifka et al., 1995; for a review, see Leslie, 2008). That is, people agree to generics for different reasons. For instance, you might agree with the statement “sharks attack swimmers”, even though only a tiny minority of sharks exhibit that property. Likewise, you might agree with the statement “ducks lay eggs”, even though most ducks (e.g., males, infertile females) can’t lay eggs.
    Generics have garnered interest among cognitive scientists, because they have the potential to provide a window into our conceptual system (Carlson, 2010; Cimpian & Markman, 2009; Gelman, 2003; Prasada, 2010). I collaborated with Sandeep Prasada, Sarah-Jane Leslie, and Sam Glucksberg (in press) to conduct a set of studies that made use of generics to analyze people’s core concepts.


    Our experiments revealed that people appear to make use of three foundational connections between kinds and properties: principled connections, causal connections, and statistical connections. The conceptual taxonomy is revealed indirectly through individuals’ assessment of generics and related generalizations.

    • Principled connections. Consider the statement, “tables are flat”. What is the connection between tables and being flat? The connection, if it exists, is not made explicit in the statement. We describe it as a principled connection, and you can test whether the connection is there by whether the following generalization sounds reasonable: “Tables, by virtue of being tables, are flat.” That is, principled connections have an explanatory dimension, whereby you can explain the property, being flat, by simply referring to the category table. Principled connections also have a normative dimension, because people appear to agree with the generalization: “Tables are supposed to be flat.” Finally, principled connections have a statistical dimension, since on average, they estimate that 88% of tables are flat.
    • Statistical connections. Some generalizations have only a statistical dimension, not an explanatory or normative one. For instance, consider the statement “fire trucks are red”. It’s a generalization that most people agree with, and on average, they estimate that 75% of fire trucks are red. Yet they appear to disagree with the statement, “Fire trucks, by virtue of being fire trucks, are red.” The statement doesn’t have an explanatory dimension, since you can’t explain the property being red by referring to the category fire truck. Likewise, people disagree with the statement, “Fire trucks are supposed to be red,” and so there doesn’t exist a normative dimension either. Such generalizations appear to have a statistical connection between the kind and property.
    • Causal connections. Why do people agree with the statement, “mosquitoes carry malaria”, even though they know that most mosquitoes don’t carry malaria? The answer is that such generalizations, known as “striking generics” (Leslie, 2008) reveal a causal connection between a kind and its property. Like statistical connections, causal connections don’t support formal explanations or normative expectations. However, they also do not support any statistical expectation that most members of the kind will have the property. Instead, what makes causal connections unique is that they suggest that the underlying makeup of a kind causes members of the kind to be disposed to have the property. For example, the biological makeup of mosquitoes causes them to be disposed to carry malaria.

    The three connections we examined form a conceptual taxonomy for generic generalization. The taxonomy has ramifications for how individuals represent and structure the world around them, how they reason about the world, and how children acquire and reason with concepts.


    • Carlson, G. (2010). Generics and concepts. In F. J. Pelletier  (Ed.), Kinds, things, and stuff. New York, NY: Oxford University Press.
    • Cimpian, A., & Markman, E. M. (2009). Information learned from generic language becomes central to children’s biological concepts: Evidence from their open-ended explanations. Cognition, 113, 14-25.
    • Gelman, S. A. (2003).  The essential child: Origins of essentialism in everyday thought.  New York:  Oxford University Press.
    • Krifka, M., Pelletier, F. J., Carlson, G., ter Meulen, A., Chierchia, G., & Link, G. (1995). Genericity: An introduction. In G. Carlson & F. J. Pelletier (Eds.), The generic book (pp. 1–125). Chicago: Chicago University Press.
    • Leslie, S. J. (2008). Generics: Cognition and acquisition. The Philosophical Review, 117, 1–49.
    • Prasada, S. (2010). Conceptual representations and some forms of genericity. In F. J. Pelletier (Ed.), Kinds, things, and stuff. New York: Oxford University Press.
    • Prasada, S., Khemlani, S., Leslie, S.-J., & Glucksberg, S. (in press). Conceptual distinctions amongst generics. Cognition.