A 10,000-dog button study moves dog cognition's evidence off owner surveys

An owner questionnaire returns what the owner concluded. A button press returns a time and a word, and presses can be counted against chance. Rossano's soundboard recordings therefore sit in a different evidence class from the owner surveys he objects to ^[1][2]. That older evidence class is part of why, for years, researchers knew little about dogs' emotions or what happens in their brains ^[3].

The reported result is that most of the 10,000 dogs used the buttons in a way far from random ^[4][5]. Scientific American's account did not report the share of dogs that met that description or the comparison the presses were tested against ^[21]. Non-random is a low bar on its own: a dog that favors the button nearest its water bowl is non-random. Some dogs used more than 100 buttons, and the article says they could effectively carry a conversation ^[6]. I would want to see the press logs before accepting that second description.

The blinking study, published last month and led by Chiara Canori, a neuroscience Ph.D. candidate at the University of Parma, took a much smaller behavior and made it the measurement ^[9][11]. Dogs blinked more when they watched other dogs blink, and they blinked back at their owners and sometimes at strangers ^[9]. The authors theorize that reciprocal blinking improves social bonds ^[10]. "Among facial expressions, there are many ways that dog use to communicate or just let us know something about their emotional state," Canori said ^[12].

An MRI study earlier this summer found that dogs' brains responded to images of happy human faces differently from sad ones ^[14]. That establishes discrimination between two classes of image, a step short of recognizing happiness, and earlier work had already shown that dogs tell words, male and female voices, barks and even different languages apart ^[15].

Rossano's project is community science, and the dogs were recorded at home on buttons that spoke words for them ^[4]. Owners remain in the pipeline as the recordists, and what gets counted is the presses. Rossano has said the same designs will be most useful when brain activity can be read in a dog's own environment ^[19].

"Compared to other animal species, dogs are significantly better at understanding human communication," Rossano said ^[7]. The field he works in is only a few decades old, and MRI on dogs is newer still ^[17]. What he wants to reach next is whether dogs know what they don't know, whether they can tell time, and whether they register events around them without reacting ^[18].