Link Search Menu Expand Document

Introduction to Commonsense Reasoning

Editors: Bill Yuchen Lin, …

Table of contents

What is commonsense reasoning?

In general, commonsense reasoning is the ability to make decisions in real-world situations with the knowledge shared by most people. Commonsense reasoning is hard to define, particularly due to the ambiguous boundary of commonsense knowledge. In the following, we show some definitions and characterizations of common sense from different authors, and finally give a summary as our working definition.

In artificial intelligence (AI), commonsense reasoning is a human-like ability to make presumptions about the type and essence of ordinary situations humans encounter every day. These assumptions include judgments about the nature of physical objects, taxonomic properties, and peoples’ intentions. A device that exhibits commonsense reasoning might be capable of drawing conclusions that are similar to humans’ folk psychology (i.e., humans’ innate ability to reason about people’s behavior and intentions) and naive physics (i.e., humans’ natural understanding of the physical world). Below are some existing definitions of commonsense knowledge.

  1. “Commonsense knowledge includes the basic facts about events (including actions) and their effects, facts about knowledge and how it is obtained, facts about beliefs and desires. It also includes the basic facts about material objects and their properties.” (McCarthy, John; 1989)
  2. “Commonsense knowledge differs from encyclopedic knowledge in that it deals with general knowledge rather than the details of specific entities.” (Tandon et al. 2018)
  3. Commonsense knowledge is “real world knowledge that can provide a basis for additional knowledge to be gathered and interpreted automatically”. (Matuszek, Cynthia, et al. 2005)
  4. The commonsense world consists of “time, space, physical interactions, people, and so on”. (Ernest Davis; Gary Marcus 2015)
  5. Common sense is “all the knowledge about the world that we take for granted but rarely state out loud”. (Clive Thompson 2018)

(Modified) Wikipedia: Commonsense Reasoning

Common sense is the basic level of practical knowledge and reasoning concerning situations and events that are commonly shared among most people. Commonsense knowledge can be categorized according to types, including but not limited to:

  • Physical common sense: Physical common sense includes knowledge about the physical properties and affordances of everyday objects.

  • Social common sense: People are capable of making inferences about other people’s mental states, e.g. what motivates them, what they are likely to do next, etc. In addition, we each have a set of social norms of accepted behavior. These are often implicit in our actions and decisions.

  • Temporal common sense: natural language rarely communicates explicit temporal information. Instead it’s vague and relies on the commonsense knowledge of the listener. This requires knowing the typical duration, times, order, frequency, etc. of events.

(Modified) A blog post based on the ACL 2020 Tutorial taught by Maarten Sap, Vered Shwartz, Antoine Bosselut, Yejin Choi, Dan Roth.

Our working definition

Commonsense reasoning is a human-like ability to make presumptions about ordinary objects, events, and situations that humans encounter every day. It can be seen a process of using commonsense knowledge to solve real-world problems, which we want to teach machines to do at the basic level. Thus, it is one of the fundamental areas of Artificial Intelligence (AI) research.

Commonsense knowledge are facts that most people can acquire in their life by observing and interacting with the physical world and other humans. Thus, there are two major topics in commonsense knowledge: physical and social common sense. Within each topic, the knowledge can be further characterized by their basic units (e.g., objects vs events) and focused dimensions (e.g., taxonomic, utility, temporal, etc.). We also discuss a few other aspects for describing commonsense knowledge.

  • Physical Commonsense Knowledge
    • objects (including both physical objects and abstract concepts)
      • taxonomic (e.g., cat is an instance of animal)
      • properties (e.g., ice is cold)
        • comparative (e.g., a book is smaller than a suitcase)
      • part-whole (e.g., birds have two legs)
      • utility (e.g., oven is used for heating something up)
      • spatial (e.g., tables and chairs located near each other)
      • affordance (e.g., open a door; put a book in a suitcase)
      • etc.
    • events
      • causality (e.g., using fossil fuel causes global warming)
      • spatial (e.g., cooking a meal happens in a kitchen)
      • duration (e.g., human’s sleeping is about eight hours)
      • order (e.g., buying a ticket happens before taking on the airplane)
      • part-whole (e.g., preparing materials is part of cooking a meal)
      • time (e.g., people usually start working at 9 AM)
      • frequency (e.g., the presidential election happens every four year in US)
      • etc.
  • Social Commonsense Knowledge
    • objects (usually abstract concepts and human activities)
      • concepts (e.g., friends help each other)
      • activities (e.g., marriage needs weddings)
      • etc.
    • events (situations in everyday life)
      • social-norm: (e.g., commenting the weight/face/age of a co-worker is not accepted)
      • motivation (e.g., X buys Y a gift because X wanted to make Y happy, …)
      • attributes (e.g., X buys Y a gift; X is seen as generous, polite, …)
      • effect (e.g., X buys Y a gift; Y is surprised)
      • etc.

There are also other dimensions to describe a commonsense fact as follows:

  • concreteness (e.g., birds have legs < birds have two legs)
  • plausibility/typicality (e.g., (most) birds have legs > (some) apples are green)
  • saliency (e.g., birds have wings > birds have legs )
  • culture-sensitiveness (e.g., high schools are four years in the USA, while three years in China.)

Bill Yuchen Lin.

Pleaser refer to the surveys and position papers for more information.

Why do we study commonsense reasoning?

How do we study commonsense reasoning?

Open Problems

Cited as (TBD)

  title   = "An Online Compendium for Commonsense Reasoning Research.",
  author  = "Lin, Bill Yuchen and Qiao, Yang and Ilievski, Filip and Zhou, Pei and Wang, Peifeng and Ren, Xiang", 
  journal = "",
  year    = "2021",
  url     = ""

Page last modified: Apr 24 2021.

Edit this page on GitHub.