"Long before it's in the papers"
January 27, 2015


Dogs may link words to object sizes rather than shapes

Nov. 28, 2012
Courtesy of the University of Lincoln
and World Science staff

Dogs re­late words to ob­jects very dif­fer­ently than hu­mans do, new re­search claims: where­as we re­late words for ob­jects pri­marily to their shapes, dogs re­late these words to sizes and tex­tures.

Many pet own­ers mar­vel at their dog’s abil­ity to fetch dif­fer­ent ob­jects such as toys on in­struc­tion, tak­ing this as ev­i­dence that the dog “un­der­stands” these words in a si­m­i­lar way to us.

Gable and toys. (Courtesy U. of LIncoln)

The new find­ings, pub­lished in the on­line re­search jour­nal PLoS One, may help to ad­vance un­der­stand­ing of the founda­t­ions of hu­man lan­guage and the crit­i­cal dif­fer­ences with oth­er spe­cies, the re­search­ers said.

Young chil­dren gen­er­al­ize names to new ob­jects on the ba­sis of shape, and con­tin­ue to do so as adult­s—a ten­den­cy known as “shape bi­as,” the sci­en­tists ex­plained. This is key to lan­guage de­vel­op­ment be­cause it en­ables chil­dren to as­sign new ob­jects to pre-established class­es—for ex­am­ple, to rec­og­nize that a ten­nis ball and a foot­ball both be­long to the cat­e­go­ry “ball.”

The re­search­ers worked with a dog and found that when he was in­tro­duced to new words to re­fer to new ob­jects, he first gen­er­al­ized based on size, then on tex­ture, but not shape.

“A num­ber of re­cent stud­ies have sug­gested that the do­mes­tic dog’s word com­pre­hen­sion is hu­man-like,” said Emile van der Zee from the Uni­vers­ity of Lin­coln in the U.K., who car­ried out the re­search with two col­leagues. Some have dis­put­ed that claim but there has­n’t been clear ex­pe­ri­men­tal ev­i­dence, he added. “Our find­ings br­ing a fun­da­men­tal new in­sight in­to this dis­cus­sion and add to our un­der­stand­ing of the cog­ni­tive equip­ment nec­es­sary for true hu­man word learn­ing.”

Van der Zee and two col­leagues worked with a five-year-old bor­der col­lie called Ga­ble who had shown re­mark­a­ble abil­i­ties to learn new ob­ject words. They de­vised four dif­fer­ent chal­lenges for Ga­ble.

On a num­ber of oc­ca­sions, a se­lec­tion of 10 dif­fer­ent ob­jects known to Ga­ble were placed in an out-of-sight en­clo­sure, and he was then giv­en a ver­bal in­struc­tion to fetch one ob­ject from the ten. In­i­tial tests con­firmed Ga­ble could easily dis­tin­guish be­tween toys he knew well. But when the re­search­ers in­tro­duced new words and nov­el ob­jects of var­y­ing shape, size and tex­ture Ga­ble be­gan to re­veal the ab­sence of shape bi­as in his choices.

He ap­peared to make dis­tinc­tions based first on ob­ject size, then, when he had long­er to be­come fa­mil­iar with the new ob­jects, on the ba­sis of tex­ture, the sci­en­tists ex­plained. Shape seemed to have no in­flu­ence. 

“This would sug­gest that an im­por­tant fac­tor in the nat­u­ral struc­tur­ing of the men­tal lex­i­con may be the way in which sen­so­ry in­forma­t­ion is or­gan­ized in a par­tic­u­lar spe­cies,” van der Zee said. “The hu­man vis­u­al sys­tem is tuned to de­tect ob­ject shape for the pur­pose of ob­ject rec­og­ni­tion. In our ex­pe­ri­ments we ex­clud­ed Ga­ble us­ing scent cues. It seems that his vis­u­al sys­tem and sen­so­ry cues linked to his mouth re­gion are fo­cused not on shape, but on size and tex­ture. Only fu­ture ex­pe­ri­ments will re­veal what role scent plays for the dog in gen­er­al­izing words. It is only by com­par­ing oth­er spe­cies with hu­mans that we can find out more about the neu­ral and ge­net­ic founda­t­ions of word re­ference in lan­guage.”

The find­ings may may al­so in­form re­fine­ments to an­i­mal train­ing pro­grams, the re­search­ers added.

* * *

Send us a comment on this story, or send it to a friend


Sign up for

On Home Page         


  • St­ar found to have lit­tle plan­ets over twice as old as our own

  • “Kind­ness curricu­lum” may bo­ost suc­cess in pre­schoolers


  • Smart­er mice with a “hum­anized” gene?

  • Was black­mail essen­tial for marr­iage to evolve?

  • Plu­to has even cold­er “twin” of sim­ilar size, studies find

  • Could simple an­ger have taught people to coop­erate?


  • F­rog said to de­scribe its home through song

  • Even r­ats will lend a help­ing paw: study

  • D­rug may undo aging-assoc­iated brain changes in ani­mals

Dogs relate words to objects very differently than humans do, new research claims: whereas people relate words for objects primarily to shapes, dogs relate these words to sizes and textures. Many pet owners marvel at their dog’s ability to fetch different objects such as toys on instruction, taking this as evidence that the dog “understands” these words in a similar way to us. Psychologists and animal behavior specialists said they showed through a series of experiments that dogs’ mental lexicon is built in a substantially different way to our own. The findings, published in the online research journal PLoS One, may help to advance understanding of the foundations of language in humans and the critical differences with other species, they added. Young children generalize names to new objects on the basis of shape, and continue to do so as adults—a tendency known as “shape bias,” the scientists explained. This is key to language development because it enables children to assign new objects to pre-established classes—for example, to recognise that a tennis ball and a football both belong to the category “ball.” The researchers found that when dogs are introduced to new words to refer to new objects, they first generalize based on size, then on texture, but not shape. “A number of recent studies have suggested that the domestic dog’s word comprehension is human-like,” said Emile van der Zee from the University of Lincoln in the U.K., who carried out the research with two colleagues. Some have disputed that claim but there hasn’t been clear experimental evidence, he added. “Our findings bring a fundamental new insight into this discussion and add to our understanding of the cognitive equipment necessary for true human word learning.” Van der Zee and two colleagues worked with a five-year-old border collie called Gable who had shown remarkable abilities to learn new object words. They devised four different challenges for Gable to determine the extent and nature of his word comprehension. On a number of occasions, a selection of 10 different objects known to Gable were placed in an out-of-sight enclosure, and he was then given a verbal instruction to fetch one object from the ten. Initial tests confirmed Gable could easily distinguish between toys he knew well. But when the researchers introduced new words and novel objects of varying shape, size and texture Gable began to reveal the absence of shape bias in his choices. He appeared to make distinctions based first on object size, then, when he had longer to become familiar with the new objects, on the basis of texture, the scientists explained. Shape seemed to have no influence. “This would suggest that an important factor in the natural structuring of the mental lexicon may be the way in which sensory information is organised in a particular species,” van der Zee said. “The human visual system is tuned to detect object shape for the purpose of object recognition. In our experiments we excluded Gable using scent cues. It seems that his visual system and sensory cues linked to his mouth region are focused not on shape, but on size and texture. Only future experiments will reveal what role scent plays for the dog in generalising words. It is only by comparing other species with humans that we can find out more about the neural and genetic foundations of word reference in language.” The findings may may also inform refinements to animal training programs, the researchers added.