Example · part of Thinking inside frameworks

Two groups read about crime in a city and saw the same statistics. For one group, crime was a “beast” preying on the city. For the other, it was a “virus” infecting it. The beast readers favored enforcement more than the virus readers did.

Thibodeau and Boroditsky ran the study in 2011 as a test of conceptual metaphor, Lakoff and Johnson’s idea that the picture used to describe a thing steers the conclusions drawn about it. The metaphor had to appear early, where it could organize the information that followed. In the book’s terms it worked like decoding instructions for the numbers.

It is a cousin of the framing effect seen in the Asian disease problem, with weaker evidence behind it. Steen and colleagues argued in 2014 that the effect was smaller than claimed and sensitive to how answers were coded. A German-language replication published in 2016 did reproduce it. The fair reading is that the effect is real, smaller and more conditional than first reported, and dependent on early placement and on what readers already believed. That places it under influence, not imprisonment.

How solid is this? Real but smaller and more conditional than first reported: contested by Steen and colleagues, partly vindicated by a German replication, and dependent on early placement and readers’ prior attitudes.

How it connects


← Conceptual metaphor · Thinking inside frameworks (15 of 21) · Theory-ladenness and paradigms →

Explore all the ideas · Read the book in order · Open the interactive map