Perceptual similarity effect in people with Down syndrome

Abstract

Background The perceptual similarity between two objects, specifically similarity in the shape of the referents, is a crucial element for relating words in earlier stages of development. The role of this perceptual similarity has not been systematically explored in children with Down syndrome (DS). Method: The aim was to explore the role of perceptual similarity in relationships between words in children with DS. Two groups, children with typical development (TD) and children with DS, matched by gender and mental age, participated in a priming task with a preferential looking paradigm. The task presented validated perceptually-related word pairs (prime-target) and perceptually unrelated pairs. In the priming task both groups were asked to look at a target image (e.g. ball) that was perceptually related (e.g. cookie) or unrelated (e.g. skirt) to the prime. Results: Participants from both groups looked more at targets without perceptual similarity than at those with similarity to the prime, suggesting an inhibition effect. Conclusions: This finding suggests the role of visual information, particularly the shape of the referents, in the construction of the lexical system.

Children with typical development (TD) begin learning new words and incorporating them into their vocabularies in their first years (Smith 2000). Various studies have investigated the linguistic elements that allow them to perceive the relationships between words that are part of their early vocabularies (Arias-Trejo and Plunkett 2013, Mani and Plunkett 2011, Styles and Plunkett 2009). Linguistic elements such as semantic information, phonological cues, and contextual concurrence are essential structural aspects of early vocabularies (Ferrand and New 2003). From the age of 12 to 24 months, children with TD attend to shape similarity (Arias-Trejo and Plunkett 2010, Hollich et al. 2007, Johnson et al. 2011, Mani et al. 2013). This preference for categorizing words according to the shape of objects, rather than other perceptual features such as color, is known as the shape bias (Graham and Poulin-Dubois 1999, Landau et al. 1988, Poulin-Dubois et al. 1999). Children with TD use shape to understand relations between words in early stages of development; in later stages they also employ other elements as cues such as texture (Arias-Trejo and Plunkett 2009, Styles and Plunkett 2009).

Researchers have posited different approaches to elucidate the association between words in development. There is a conceptual dimension, such as words that belong to a common semantic category, like cat and dog (Lucas 2000). There is also a non-conceptual dimension (Mani et al. 2013) that includes observable features such as shape and color, as in the association between moon and ball, which belong to different semantic categories but share the perceptual quality of shape. Recent studies of children with TD have used visual preference tasks to show that they employ the perceptual similarity of images to understand the relationships of words (Arias-Trejo and Plunkett 2010, Johnson et al. 2011, Mani et al. 2013). Such studies are lacking, however, for people with Down syndrome (DS), but there are studies comparing the use of conceptual and perceptual information in people with autism (Hetzroni et al. 2019).

Trisomy 21 (Down syndrome (DS)) is the most common genetic cause of intellectual disability (Nadal and Estivill 2001), with an international incidence of 1 per 1,499 live births (Graaf et al. 2017). It is characterized by a dissociation between chronological age (CA) and mental age (MA), a dissociation that tends to increase with CA (Facon et al. 1998). Many studies of DS have established a specific cognitive phenotype with notable deficits in the area of language, especially in oral production (Abbeduto et al. 2007, Arias-Trejo and Barrón-Martínez 2017, Chapman 1997, Chapman et al. 1991, Chapman et al. 1998), word learning (Rondal 1995), and in executive functions such as inhibition, planning processes, and cognitive flexibility (Lanfranchi et al. 2010, Rowe et al. 2006). There is, however, no consensus about strengths or deficits in perceptual ability (Saviolo-Negrin et al. 2005, Wan et al. 2015), the ability to process visual information that involves spatial relations or other features of objects. Several studies have focused on investigating domains of perceptual ability in people with DS. For example, Wan et al. (2015) compared the visual perceptual ability of people with DS and chronologically-matched peers with TD with an average age of 11 years in a study with 70 participants in each group. Their study used the Test of Visual Perceptual Skill (TVPS; Martin and Gardner 2006), which has subtests including visual discrimination, spatial relationships, and form constancy. The typical group showed better performance (fewer errors) on all subtests than the DS group. This means that perceptual ability is significantly reduced in the DS group, specifically regarding visual and sequential memory and visuospatial relationships. In addition, correlational analysis showed that performance in both groups was positively related to chronological age (range: 6-20 years): that is, visual perceptual functions improved with age.

Another study (Saviolo-Negrin et al. 2005) evaluated the visual perceptual ability of people with DS with a chronological age of 14 to 43 years, using the Frostig Development Test of Visual Perception (Frostig et al. 1963), with subtests for spatial relationships, form constancy, figure-ground, and space position. The results were analyzed in two chronological subgroups, 14-25 years (average MA 5.25 years) and 25-43 years (average MA 4.85 years), to explore the aging effect. The younger group showed better performance (fewer errors) than the older group on all subtests, and correlational analyses revealed a negative correlation between subtest scores and chronological age. That is, visual perceptual ability decreased with an increase in chronological age, specifically in relation to figure-ground and rotation of objects.

Along the same lines, Yang et al. (2014) have asserted, based on an extensive review of empirical studies, that perceptual skills are not generally a cognitive strength in individuals with DS. They argue that some perceptual skills, such as spatial sequential memory, working memory, and mental rotation, are similar to those of their peers with the same MA, while others, such as visuospatial construction and Gestalt closure, are superior in those with typical development. This finding suggests that the variation in perceptual abilities in people with DS may be due to various factors: 1) the type of perceptual skill, since there are a variety of perceptual subdomains (Lohman et al. 1987); 2) the cognitive ability linked to the perceptual skill, such as memory (Godfrey and Raitano Lee 2018) or motor development (Vicari 2006); 3) the methodology used, since the instruments are different and do not allow for direct comparison of the results (Yang et al. 2014); and 4) age, since some studies examine young people and others focus on adults without considering the differences associated with aging in CA and MA. To our knowledge, there have been no studies focusing on the perceptual ability linked to the discrimination of objects that relates to lexical organization, including the shape bias, in people with DS.

The perceptual ability of people with DS has been evaluated mainly with standardized tests linked to processes like working or visual memory, rotation of objects, identification of patterns, categorization, and motor coordination (Saviolo-Negrin et al. 2005, Wan et al. 2015). There have been no studies that specifically and directly evaluate perceptual ability as an element of word association in persons with DS, without the presence of explicit memory processes. The ability to link words has traditionally been measured with adults with TD using priming tasks, where the recognition of a target word is more efficient when it is preceded by a related than an unrelated prime (Meyer and Schvaneveldt 1971). The methodological advantage of using a priming task in a preferential looking paradigm over a neuropsychological test is that it excludes explicit verbal ability and working memory, and its processing is relatively effortless. The preferential looking methodology has previously been employed in children with TD to study perceptual similarity from 12 to 24 months (Arias-Trejo and Plunkett 2010, Hollich et al. 2007, Johnson et al. 2011, Mani et al. 2013). Children with DS should be studied at a similar MA, meaning a more advanced CA than those with TD, in order to explore their perceptual processes. Wiggs and Martin (1998) describe the mechanisms and properties of priming linked to perceptual similarity: it is independent of explicit tasks or tests, it does not require working memory, it is manifested in short time windows given its automaticity effect, it is not affected by characteristics of presentation such as rotation, position, depth, or texture, and it is obtained with simple representations of fundamental features.

The main objective of our study was to evaluate perceptual similarity as an element that allows the establishment of links between two words (e.g. ball-moon), using a preferential looking priming task, that is, to determine whether children with DS showed a different visual pattern when exposed to prime and target words whose referents were perceptually similar than when they were dissimilar. The results would provide evidence about the perceptual elements that allow people with DS to link words in the absence of conceptual elements. A group of children with TD, matched by MA and gender, was also evaluated to compare with the results from the experimental DS group. The research protocol was reviewed and approved by the ethical review committee of the UNAM Department of Postgraduate Studies of the Faculty of Psychology.

Method

Participants

Two groups of 25 children each were included in a cross-sectional study: an experimental group with Down syndrome (DS; 14 boys and 11 girls) and a control group with typical development (TD) matched by mental age (MA) and gender. The DS group had a mean MA of 3.76 years (SD = 1.29, range: 3.08-4.75 years) and a mean chronological age (CA) of 10.74 years (SD = 6.17, range: 5.75-16.83); the TD group had a mean MA of 3.92 years (SD = 1.36, range: 3.25-5.01) and a mean CA of 3.81 years (SD = 1.32, range: 3.08-4.95). From initial samples of 29 children with DS and 26 with TD, four children with DS were excluded because they failed to pay attention in more than 50% of the trials (n = 2) or did not fulfill the inclusion criteria (n = 2), and one child with TD was excluded for lack of attention.

Children were recruited from public and private day care centers and through advertisements published in the university gazette and the public transportation system. All participants resided in Mexico City and the surrounding metropolitan area, and all were monolingual in Mexican Spanish. The inclusion criteria for both groups were: no visual or hearing problems (as verified by otoacoustic emissions audiometry and a standardized visual acuity test, both applied by trained laboratory staff), no neurological or psychiatric problems (according to parental report), and regular trisomy (T21) in the case of children with DS. The children with DS and children with TD were matched by MA using a reliable abbreviated version (Sattler 2010) of the Wechsler Preschool and Primary Scale of Intelligence-III (Wechsler 2003), made up of three subscales: Block Design (the child views a model and/or a picture and is asked to use one or two colored blocks to recreate the design); Receptive Vocabulary (the child is asked to point to one image of four presented by the examiner); and Object Assembly (the child assembles the pieces of different puzzles to create a representation of an identified object until she fails). The reliability and validity values of the abbreviated version are 0.93 and 0.74, respectively.

Experimental task

Stimuli

The task used 30 concrete and familiar nouns, all of early acquisition according to the MacArthur-Bates Communicative Development Inventories for Spanish-speaking Children with Down Syndrome (Galeote et al. 2006). Of the 30, 10 served as primes, 10 as targets, and 10 as distractors. The prime and target words were presented orally in each trial. Target and distractor images were organized in such a way as to form fixed pairs. Two conditions were created according to the relationship between the prime and the target: Perceptually Related and Perceptually Unrelated. In the Related condition, the prime and target words shared a perceptual relation determined by their shape or contour and no semantic, phonological, or other lexical relationship (as explained further in the section below on perceptual validation). Pairs in the Unrelated condition had no perceptual, semantic, or phonological association. The unrelated pairs were created by changing the position of the prime in the fixed target-distractor pairs with the objective of eliminating the perceptual relationship between the prime and the target. Table 1 shows the prime and the target-distractor pairs. The distractor did not have a perceptual relationship with the prime or target (or a phonological, semantic, or associative relationship) in either condition.

Table 1.

Primes, targets, and distractors used in perceptual related and unrelated conditions.

Perceptual Related Condition			Perceptual Unrelated Condition
Prime	Target	Distractor	Prime	Target	Distractor
tortilla [tortilla]	reloj [clock]	bicicleta [bicycle]	plátano [banana]	reloj [clock]	bicicleta [bicycle]
pantalón [pants]	escalera [stairs]	mariposa [butterfly]	galleta [cookie]	escalera [stairs]	mariposa [butterfly]
lápiz [pencil]	flecha [arrow]	playera [t-shirt]	tortilla [tortilla]	flecha [arrow]	playera [t-shirt]
plátano [banana]	teléfono [phone]	conejo [rabbit]	pantalón [pants]	teléfono [phone]	conejo [rabbit]
galleta [cookie]	pelota [ball]	tortuga [tortoise]	lápiz [pencil]	pelota [ball]	tortuga [tortoise]
piano [piano]	mesa [table]	serpiente [snake]	paleta [lollipop]	mesa [table]	serpiente [snake]
gato [cat]	florero [vase]	limón [lime]	jirafa [giraffe]	florero [vase]	limón [lime]
paleta [lollipop]	árbol [tree]	esponja [sponge]	rueda [wheel]	árbol [tree]	esponja [sponge]
jirafa [giraffe]	lámpara [lamp]	avión [airplane]	piano [piano]	lámpara [lamp]	avión [airplane]
rueda [wheel]	luna [moon]	helicóptero [helicopter]	gato [cat]	luna [moon]	helicóptero [helicopter]

Auditory stimuli

All the nouns employed were digitally recorded by a female Spanish speaker, using infant-direct speech (Cooper and Aslin 1990), in a soundproof room; they were edited at 44,100 Hz and 16-bits and normalized and adjusted in amplitude and volume.

Visual stimuli

Visual stimuli were 10 target-distractor image pairs selected from open-source image databases. All target-distractor image pairs were of the same size (1,440 x 1,080 pixels) and were presented on a gray background. The 20 color images employed were concrete, familiar, and prototypical, validated in a previous study.

Perceptual validation study

In order to corroborate that the prime and the target in the related condition shared a perceptual relation by shape, an experimental validation was made with a different group of participants than those evaluated in the experiment. The results of this validation support the pairs of words used in the experimental task. The validation involved 20 schoolchildren with TD (M = 11.79 years, SD = 0.25, range: 11.25-12.25 years) from a public elementary school in Mexico City. The validation of stimuli was carried out with PsychoPy software (version 1.82.01), which allowed the presentation of visual stimuli and storage of answers. The experiment consisted of 24 trials of two types: 12 perceptually related (e.g. cookie-ball) and 12 perceptually unrelated (e.g. cookie-skirt), which were shown to the participants in a randomized manner. In each trial, two images were shown simultaneously for a period of 10 s. The task of the participant was to indicate, using the computer mouse, the similarity in shape of the two images on a scale of 1 to 5. A score of 1 indicated that the images were not similar in shape (e.g. cat-moon), and 5 indicated that they were similar in shape (e.g. ball-moon). Instructions were provided to participants in written and oral form before beginning the task to avoid possible comprehension difficulties in children with reading deficits. If the participant did not generate a response within 10 s, the system automatically continued to the next trial; the participant could not return to a previous trial. The total duration of the task was less than five minutes. For data analysis, the mean was obtained of the scores selected for each pair of images. The scores of two children who did not complete at least 80% of the trials were excluded from the analysis. Pairs of images with scores above the median (Mdn = 3) were considered perceptually related. The results show that of the 12 perceptually related pairs presented, 10 had scores greater than 3 (M = 4.32, SD = 0.83, range: 3.05-4.85, variance = 0.28), and these were used for the perceptually related trials (e.g. banana-phone). Of the 12 perceptually unrelated pairs, 11 had scores below the median (M = 1.42, SD = 0.62, range: 1.05-2.30, variance = 0.11), and were used as distractors (e.g. banana-horse).

Experimental design

Ten trials were presented, five for each condition (Perceptually Related and Perceptually Unrelated), in pseudo-randomized order, so that no more than two trials from the same condition appeared consecutively. Each trial had a total duration of 4,700 ms. Images appeared on the left and right side of the monitor and were used the same number of times as targets or distractors. The inter-stimulus interval (ISI) and the stimulus onset asynchrony (SOA) were both 200 ms. Each trial was conducted according to the following sequence: an attention getter (0 to 500 ms), the auditory prime word (2,000 ms), the ISI (2,000 to 2,200 ms), target and distractor images appearing simultaneously (2,200 to 4,700 ms), SOA (2,200 to 2,400 ms), and finally the auditory target word (2,400 ms). Figure 1 illustrates this sequence.

Figure 1.

Example of sequence of a Perceptual Related trial.

Experimental measure

A portable Tobii X2-30 eye-tracker was employed to monitor participants’ visual attention. This device records and analyzes eye movements using infrared diodes to generate reflection patterns in participants’ corneas and capture visual preferences. Reflection data are collected by sensors; the device calculates the three-dimensional position of each eyeball and finally the point on the screen at which the person is looking.

The data obtained in the visual tracking task were automatically coded by the Tobii studio software, which yielded the number of visual fixations of both eyes by each participant during the duration of the 10 trials presented. Trials were eliminated in which the participants were not looking or in which the eye-tracker had calibration problems as indicated by the Tobii software). The sampling rate of the Tobii software was 30 Hz. The measure employed for the statistical analysis was the Proportion of Target Looking (PTL), the duration of attention to the target divided by the total duration of attention to the target and distractor images (T/[T + D]). This measure was employed for each participant in each type of trial, and at the end it was averaged by group. For example, if T = 300 ms and D = 700 ms, applying the formula, (300/[300 + 700]) = 0.30. Cleaning criteria were employed for data analysis, as explained in the Results section.

Procedure

Sociodemographic information to validate inclusion criteria and written informed consent, signed by the primary caregiver, were obtained before the experimental session. Participants in both groups were evaluated in a single session in the laboratory or at their school or day care, depending on parental availability. First, we evaluated the hearing and vision of the participants through otoacoustic emissions audiometry and a standardized visual acuity test, both applied by trained laboratory staff (approximately 10 min). Then, participants’ MA was measured using a reliable abbreviated version (Sattler 2010) of the Wechsler Preschool and Primary Scale of Intelligence-III (WPPSI-III, Wechsler 2003), whose application time was approximately 30 min without rest periods. During the WPPI-III evaluation the Block, Receptive Vocabulary, and Object Assembly subtests were applied in random order. The experimental priming task was applied next (approximately 8 min), with each child seated at a distance of 60 cm from an LED monitor (23-inch, resolution 1,920 x 1,080). Participants were instructed to look at the images on the monitor and to avoid moving their face or body, pointing at the monitor, or talking. A five-point calibration video with a cartoon image attention-getter was presented to each participant at the start of the experimental session. The experimental task began when at least three of the five points were successfully calibrated for both eyes. Two familiarization trials with two cartoon images were presented in order to ensure the participants’ attention, followed by the 10 experimental trials. The experimenter managed the presentation of the experiment without interacting with the participants. The total session lasted approximately one hour and the performance results of each participant were shared with their parents or guardians in written reports.

Results

A comparison of means test between mental age (MA) and chronological age (CA) was performed in both groups with the Student t-test. In the TD group, as expected, no statistically significant differences were found between MA and CA (t (24) = −2.58, p = 0.82). In the group with DS, a statistically significant gap was found between MA and CA, as in previous studies (t (24) = 7.78, p = 0.00). In addition, no differences were found between the MA of the group with DS and the MA of the group with TD (t (24) = −7.80, p = 0.73). As expected, differences were found in CA between the DS and TD groups (t (24) = −7.62, p = 0.00). The Proportion of Target Looking (PTL) was employed in the analysis of the experimental task.

Four cleaning criteria were employed for data analysis, eliminating the following: 1) trials with no attention; 2) trials with less than 10% attention to the target and distractor within the analysis window (less than 200 ms); 3) participants who completed less than 60% of the trials (three Related and three Unrelated); and 4) outliers of more than two standard deviations, which were replaced by the mean for all participants. Of the 500 original trials with both groups (ten trials each for 50 participants, 25 from each group), 425 (85%) were analyzed: 205 from the DS group and 220 from the TD group. The analysis was performed for the interval window from 2,700 to 4,700 ms, that is, beginning 300 ms after the auditory target word was heard, because a previous study (Canfield et al. 1997) suggests that a looking task requires a short reaction time to evoke a participant’s visual response. The data were analyzed in a 2 × 2 repeated measures analysis of variance (ANOVA) with the factors Condition (Perceptual Related or Unrelated) as a within-subjects factor and Group (Down Syndrome or Typical Development) as a between-subjects factor. The results showed a significant effect of Condition (F (1,48) = 10.95, p = .001, ɳ² = .18). The variance was 0.016 (SD = 0.12) for Perceptual Related trials and 0.015 (SD = 0.12) for Perceptual Unrelated trials. Both DS and TD groups showed a significantly greater PTL in Perceptual Unrelated than in Related trials. There were no other significant effects or interactions. The results are shown in Figure 2; for illustrative purposes, the mean PTL is separated by group.

Figure 2.

Mean Proportion of Target Looking (PTL) in Related and Unrelated Trials in Down Syndrome and Typical Development groups. The horizontal line represents chance level (+/− SE).

In order to analyze the response obtained in two time windows, we performed comparisons between Perceptual Related and Unrelated trials in two time subwindows of 1,000 ms in the two groups: subwindow 1, from 2,700 to 3,700 ms, and subwindow 2, from 3,701 to 4,700 ms. In the case of the TD group, an inhibition effect was found in subwindow 1 but not in subwindow 2. In subwindow 1 there was a significantly greater proportion of attention to Perceptual Unrelated than to Related trials (t (29) = −4.53, p = 0.00, d = −0.54). However, in window 2 there were no significant differences between the trials (t (25) = −0.80, p = 0.43). In the TD group the inhibition effect appeared early and faded toward the end of the trial. In the DS group, an inhibition effect was found in both subwindows 1 and 2. There was a greater proportion of attention to Perceptual Unrelated than to Related trials, both for subwindow 1 (t (25) = −3.62, p = 0.00, d = −0.49) and for subwindow 2 (t (28) = −3.53, p = 0.00, d = −0.39).

Discussion

The aim of this study was to use a priming task in a preferential looking paradigm to explore the role of perceptual similarity as an element that allows people with Down syndrome (DS) to establish links based on shape between two words (e.g. ball-moon). It makes a methodological contribution in using visual tracking as a new technique to measure lexical organization through perceptual similarity. The results showed significant differences in the proportion of attention in each experimental condition (Perceptually Related vs. Unrelated) between the two groups. Participants in both DS and TD groups showed less target attention in the trials with perceptual similarity (e.g. ball-moon) than in those without perceptual similarity (e.g. cat-moon). The perceptual similarity by shape was duly validated by the perceptual validation study described in the Method section. This pattern is known in similar studies of priming as an inhibition effect (Goldiner et al. 1989, Mani et al. 2012) and supposes an effect of lexical interference and longer attention in unrelated than in related trials. In a conventional priming task, we would expect greater attention in the related trials, which would indicate a lexical facilitation effect. However, we found the opposite pattern, suggesting an inhibitory effect in which the common perceptual elements between two objects evoke longer attention times (Linck et al. 2008). In addition, the statistical analysis in two time windows shows an inhibition effect that is more prolonged in the DS group than in the TD group, a difference that is probably related to the perceptual processing strategies of each group.

Our study sheds light on this inhibitory effect by providing the first experimental evaluation in people with DS of the use of perceptual (non-conceptual) skills to identify the relationship between two words. Prior studies of the subject (Saviolo-Negrin et al. 2005, Wan et al. 2015) evaluated the perceptual ability of children and adults with DS focusing only on conceptual features of the objects, utilizing standardized tests to measure skills such as identification of patterns, categorization, and motor coordination: abilities more linked to motor functions. Taking into consideration the pattern of results found in this study, we assume that the inhibitory effect found in the priming task reflects a greater cognitive load, since the participant not only generates a prototypical mental candidate of the prime word that is heard, but also discriminates between target and distractor images shown simultaneously, one perceptually similar to the prime and the other not.

The lack of differences between the DS and TD groups could be explained by the use of similar mechanisms or cognitive strategies to detect the perceptual information in two different referents. It is probable that in both groups the absence of perceptual similarity evokes an interference effect that demands more attention to perceptually dissimilar pairs of words than to similar ones: a search (reflected in increased attention) for common perceptual elements between the word that is heard and the referents of the images. It is worth noting that, in the present study, the same pattern of results was found in children with TD at 4 years of MA.

Wiggs and Martin (1998) describe various properties or mechanisms involved in perceptual priming. Some of these rely on an automaticity effect and do not require working memory. Perceptual priming is also not affected by the rotation, position, depth, or texture with which objects are presented, but by the simple representation of the object. The time window employed in the present experiment (2,000 ms) was appropriate for the analysis of differences between the perceptually related and unrelated trials in both groups, since the mechanism involved in perceptual priming operates without the influence of working memory. In addition, Gentner (1983, 2010) has proposed a systematicity effect to explain how patterns of similar features (for example, the position of objects or their profile) allow for optimal identification of referents.

The inhibition effect shown by children with DS implies that like their typical counterparts, they possess the necessary cognitive mechanisms to allow differentiated attention to perceptual phenomena in tasks related to visual and auditory stimuli. Lexical inhibition allows people to prioritize their attention to different elements according to the cues that are provided. The inhibition effect is a superior psychological skill, part of an individual’s executive functioning that is related to planning processes and cognitive flexibility (Lanfranchi et al. 2010, Rowe et al. 2006). Importantly, despite the difficulties in lexical production in people with DS (Abbeduto et al. 2007, Chapman 1997), they appear to show no deficit in this effect. This finding gives us valuable information about the perceptual elements visually taken into account by people with DS (as well as their MA peers) to structure their system of lexical relationships. The shape of objects plays an important role in their lexical organization.

Our results also confirm the existence of the theoretical phenomenon of the shape bias (Landau et al. 1988) in people with DS, using an experimental priming task with perceptually similar and dissimilar stimuli, which had not been previously demonstrated. Recent studies (e.g. Tovar et al. 2019) have found no evidence of the shape bias in populations with other intellectual disabilities, such as autism, and future research should investigate what variables allow perceptual similarity to affect the recognition of words in two or more atypical populations. Such comparisons could, for example, focus on exploring the influence of vocabulary on the development of the shape bias. The results of the present study point to a syndromic specificity in DS in terms of perceptual similarity and lexical organization.

The results of our study may have implications for word learning by people with DS (Rondal 1995). Our results suggest the utility of highlighting the common perceptual aspects of words taught to children with DS. We now have lexical keys that can be used to teach words, at least on an experimental level, using nouns whose referents share perceptual features. If we know that people with DS can process characteristics that are common or not common to two objects, we can teach them more complex associations and enrich their vocabulary. Teaching based on the shape of objects could expand the categories that children learn from an early age. In addition, if we promote principles of word learning such as mutual exclusivity (Markman and Wachtel 1988), we can match common characteristics to help children with DS infer the meaning of new words based on object-shape information. In this way, and without explicit instructions, children with DS could acquire new vocabulary without resorting to the conceptual definition of the new word, which would represent an advantage to their lexical development. It would also be interesting to consider the combination of lexical (e.g. phonological) categories (Mengoni et al. 2014) in word learning, in order to expand and enrich the lexical network of people with DS. There are few word teaching programs for people with DS based on empirical research. Our findings could be used in the design of intervention tasks to improve and expand the comprehensive and oral vocabulary of the population with DS.

One limitation of our study is that the perceptual validation study was originally designed for and carried out in a younger population than the one in this study, and one with typical development. However, it provided us with a strategy to guarantee that the perceptual relationships used in the experiment could be identified as similar in shape by participants in later stages of development but with similar chronological age and experience.

Finally, additional research is necessary to explore the role of other perceptual elements in people with DS, such as color, or elements that combine perceptual similarity with characteristics such as semantic category, as has been done for children with TD (Arias-Trejo and Plunkett 2010).

Funding Statement

This work was supported with a doctoral scholarship by the Consejo Nacional de Ciencia y Tecnología, Programa de Maestría y Doctorado en Psicología, UNAM [CVU 620299], and Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica [IN304417] awarded to Natalia Arias-Trejo.

Disclosure statement

No potential conflict of interest was reported by the authors.