Absolute Numerosity Discrimination as a Case Study in Comparative Vertebrate Intelligence

Abstract

The question of whether some non-human animal species are more intelligent than others is a reoccurring theme in comparative psychology. To convincingly address this question, exact comparability of behavioral methodology and data across species is required. The current article explores one of the rare cases in which three vertebrate species (humans, macaques, and crows) experienced identical experimental conditions during the investigation of a core cognitive capability - the abstract categorization of absolute numerical quantity. We found that not every vertebrate species studied in numerical cognition were able to flexibly discriminate absolute numerosity, which suggests qualitative differences in numerical intelligence are present between vertebrates. Additionally, systematic differences in numerosity judgment accuracy exist among those species that could master abstract and flexible judgments of absolute numerosity, thus arguing for quantitative differences between vertebrates. These results demonstrate that Macphail's Null Hypotheses - which suggests that all non-human vertebrates are qualitatively and quantitatively of equal intelligence - is untenable.

Keywords: categorication; crow; intelligence; monkey (Macaca mulatta); number cognition.

FIGURE 1

Discrimination performance for simultaneously presented small numerosities. (A) Layout of the delayed match-to-numerosity task (DMNT) for dot arrays. (B) Example stimulus protocols for numerosity 1–5 that control for different non-numerical parameters. (C) Average numerosity performance functions of two rhesus macaques in the DMNT for target numerosity 2–6 (data from Nieder and Miller, 2003). (D) Average numerosity performance functions of two carrion crows in the DMNT for target numerosity 1–5 (data from Ditz and Nieder, 2015). (E) Weber fractions for small simultaneous-numerosity discriminations of two macaques and two crows. Weber fractions derived from the functions shown in (C,D), respectively.

FIGURE 2

Ideal numerosity performance function. (A) Ideal numerosity performance function for target numerosity 10 plotted on a linear number scale (top graph). The function shows a steeper slope toward smaller, and a shallower slope toward higher numerosity. As a result, the just-noticeable difference (JND, indicated by dotted colored lines) at which numerosities smaller (n_S) and larger numerosities (n_L) can be discriminated in 50% from the target (n) is smaller on the left compared to the right side of the function. (B) When the same function is plotted on a logarithmic number scale, the function becomes symmetric and the JNDs are equal on either side of the function (bottom graph).

FIGURE 3

Discrimination performance for simultaneously presented large numerosities. (A) Average numerosity performance functions of two carrion crows in the delayed match-to-numerosity task (DMNT) for target numerosity 1–30 (data from Ditz and Nieder, 2016). (B) Average numerosity performance functions of two rhesus macaques in the DMNT for target numerosity 1–30 (data from Merten and Nieder, 2009). (C) Average numerosity performance functions of 20 humans in the DMNT for target numerosity 1–30 (data from Merten and Nieder, 2009). (D) Weber fractions for large simultaneous-numerosity discriminations of two crows, two macaques, and 20 humans. Weber fractions derived from the functions shown in (A–C), respectively.

FIGURE 4

Discrimination performance for sequentially presented numerosity. (A) Layout of the delayed match-to-numerosity task (DMNT) for four sequentially presented single dot in the sample period. (B) Average numerosity performance functions of two rhesus macaques in the sequential DMNT for target numerosity 1–4 (data from Nieder, 2012). (C) Average numerosity performance functions of two carrion crows in the sequential DMNT for target numerosity 1–4 (data from Ditz and Nieder, 2020). (D) Weber fractions for small sequential-numerosity discriminations of two macaques and two crows. Weber fractions derived from the functions shown in (B,C), respectively.