Printer Friendly

Responding and learning by exclusion in 2-year-olds: the case of adjectives.

The presence of a novel stimulus amidst other familiar stimuli is one condition that can potentially facilitate word learning (Carey & Bartlett, 1978; Vincent-Smith, Bricker, & Bricker, 1974). Over the past 40 years, learning by exclusion (also known as fast mapping or emergent symbolic matching) has been studied by researchers in various areas, and particularly by developmental psychologists interested in language development (e.g., Carey & Bartlett, 1978; Clark, 1983; 1987; Diesendruck, 2005; Golinkoff, Hirsh-Pasek, Bailey, & Wenger, 1992; Hall & Graham, 1999; Markman, 1987; 1989; Markman, & Hutchinson, 1984; Markman & Wachtel, 1988; Merriman & Bowman, 1989; Swingley, 2010), and behavior analysts (e.g., Carr, 2003; Cipolla, Domeniconi, & Machado, 2014; Costa, Wilkinson, McIlvane, & de Souza, 2001; Costa, Grisante, Domeniconi, de Rose, & de Souza, 2013; Dixon, 1977; Greer & Du, 2015; McIlvane, et al., 1987; McIlvane, Kledaras, Lowry, & Stoddard, 1992; McIlvane & Stoddard, 1981; 1985; Stromer, 1986; 1989; Wilkinson & McIlvane, 1994; 1997).

Responding by exclusion is usually investigated in the context of a baseline of auditory-visual conditional discriminations for which auditory samples are names (i.e., word-object relations). On exclusion probes, the experimenter presents one novel (undefined) dictated name, and an array of visual stimuli (one undefined visual stimulus and one or more defined comparisons); under a variety of conditions, participants typically select the unfamiliar comparison (emergent responding).

In one of the initial studies on exclusion responding, Vincent-Smith et al. (1974) reported that, after taking up to five sessions to learn the relation between a novel word and one of two objects presented simultaneously, children aged 20 to 31 months quickly learned the relation between another novel word and a second object. Based on these findings, the authors argued that the use of at least one familiar stimulus could facilitate the learning of new symbolic relations. Dixon (1977) followed that suggestion by teaching individuals with intellectual disabilities to select visual stimuli under the control of spoken words. During a sequence of auditory-visual matching-to-sample (MTS) trials, the experimenter presented two printed Greek letters (pi and theta, or pi and upsilon) as simultaneous comparison stimuli. In the initial phase, all trials included the same spoken word /pi/ as the sample stimulus. After the participant consistently selected the letter pi ([pi]), the experimenter introduced probe trials, interspersed among training trials, in which the sample stimulus was either /theta/ or /upsilon/. All participants immediately selected the corresponding printed letter on those probe trials. According to Dixon, in order to do that, participants had to reject the letter pi (related to the word /pi/); that is, the controlling relation of the novel word over the selection of a stimulus never selected before was, in fact, the rejection of the visual stimulus related to the word /pi/.

These initial findings on the exclusion effect have been systematically replicated with hundreds of human participants in different laboratories: typically developing children, children with developmental or intellectual disabilities, and undergraduate students (e.g., Carr, 2003; Cipolla et al., 2014; Costa, 2009; Costa, de Rose, & de Souza, 2010; Costa et al, 2001; Costa et al., 2013; Dixon, 1977; Dixon, Dixon, & Spradlin, 1983; Domeniconi, Costa, de Souza, & de Rose, 2007; Ferrari, de Rose, & McIlvane, 1993; Greer & Du, 2015; McIlvane, Bass, O' Brian, Gerovac, & Stoddard, 1984; McIlvane et al., 1987; 1992; McIlvane, Munson, & Stoddard, 1988; McIlvane & Stoddard, 1981; 1985; McIlvane, Withstandley, & Stoddard, 1984; Sidman, 1985; Stromer, 1986; 1989; Stromer & Osborne, 1982; Wilkinson, Dube, & McIlvane; 1998; Wilkinson & McIlvane, 1994; 1997). Exclusion responding has also been demonstrated with nonhuman subjects in simple and conditional discriminations (e.g., Beran & Washbum, 2002; Clement & Zentall, 2003; Herman, Richards, & Wolz, 1984; Kaminski, Call, & Fisher 2004; Kastak & Schusterman 2002; de Souza & Schmidt, 2014; Tomonaga, 1993), which suggests that exclusion responding may indeed be a basic behavioral process (Wilkinson, de Souza, & McIlvane, 2000).

Additional evidence comes from studies using MTS procedures, during which a participant may select the correct comparison stimulus based either on the rejection of an S-or on the selection of the S+ (Johnson & Sidman, 1993; McIlvane et al., 1987). Wilkinson & McIlvane (1997), for example, showed that these two routes of stimulus control might occur concomitantly, which seems to explain why exclusion is such a robust phenomenon (Wilkinson et al., 2000). In Wilkinson and McIlvane's (1997) procedure, each trial presented an auditory sample stimulus and three visual stimuli as comparisons. On each trial, one of the comparisons was a "blank" comparison (termed the "mask"), that is, a gray square, that covered one of the comparison stimuli. Throughout training, the mask always covered one of the comparisons (either the S+ or one of the two S- stimuli). The selection of the mask was programmed as correct for half of the trials. By correctly selecting a picture comparison or the mask, participants engaged in typical "Yes'7 "No" responding, thus indicating on each trial whether the correct comparison was one of the pictures. Probes inserted among baseline trials showed that responding occurred under the control of both sample/S+ (selection) and sample/S- (rejection) relations. Therefore, "blank comparison methods also allow one to determine how children respond to word learning probes, particularly those posed by single-trial exposures" (Wilkinson & McIlvane, 1997, p. 128).

Despite the extensive literature on exclusion responding, some issues still require further investigation, such as generality. Results have been extended to different populations, but the majority of empirical studies on exclusion responding with human participants have addressed word-object relations. Although the literature on word learning suggests that children learn object names before other word classes (Gentner, 1982; Golinkoff, Mervis, & Hirsh-Pasek, 1994; Poulin-Dubois, & Graham, 2007), it remains controversial whether these results reflect different rates of acquisition or different verbal communities' practices (Waxman et al., 2013; see also Choi, 2000 and Tardif, 1996, for word learning in Mandarin and Korean, respectively).

Therefore, research on exclusion involving other word (lexical) classes is timely. From the perspective of developmental psychologists and psycholinguists, the main question on word acquisition concerns how young children learn words (nouns, verbs, adjectives, etc.). Behavior analysts, on the other hand, would rather emphasize that the child learns operant verbal behaviors (e.g., tacts, mands, intraverbals; cf. Skinner, 1957), under the control of relations between stimuli (spoken words and different aspects of the environment): word-object, word-action, and word-property. Exclusion responding involving these types of stimulus-stimulus relations has been extensively demonstrated, but the focus of research has been name-object relations, whereas few studies have investigated word-property relations (Carey & Bartlett 1978; Costa, 2009; Costa et al., 2010; Heibeck & Markman, 1987; Holland, Simpson, & Riggs, 2015). Holland et al., for example, reported that 3 and 4-year-old children showed similarly high accuracy rates in selecting an undefined object (displayed simultaneously with two familiar objects) under the control of different types of words (object, shape, color, and texture labels). Using a within-subjects design, Costa (2009) investigated word-object, word-property, and word-action relations. Sixty participants (children aged 24 to 36 months) were exposed to these three conditions in a counterbalanced order. The procedure involved a three-choice MTS task, and used the blank comparison procedure (Wilkinson & McIlvane, 1997), in which a gray mask covered one of the comparison stimuli (the S+ on half of the trials, and the S- on the other half of trials). All the participants selected the undefined comparison stimulus on exclusion probes, and the overall percentage of responding was 100% for name-object relations, 98% for word-action relations, and 93% for word-property relations. Participants who did not select the novel comparison usually selected the mask, but never a defined stimulus. Younger participants required more training to achieve criterion in baseline and showed less responding under the control of novel words when dictated words were adjectives or verbs than when they were nouns.

Based on Costa's findings (2009), the present study investigated exclusion responding in young children (24- to 29month-olds) by introducing exclusion trials interspersed among baseline trials (auditory-visual conditional discriminations), during which sample stimuli corresponded either to objects (nouns) or properties (adjectives). Additionally, although exclusion responding has been extensively replicated, a relevant issue is whether the selection of an undefined comparison in the presence of an undefined sample results in learning the relation between those two stimuli (i.e., the emergence of a conditional discrimination) on exclusion trials. A variety of learning outcome tests have been designed to evaluate what is learned on exclusion trials (Wilkinson & McIlvane, 1997; Wilkinson, et al., 1998).

After experiencing just one exclusion trial, participants from different studies have shown responding consistent with having learned the sample/comparison relations on some types of learning outcome probes, but a high percentage of inconsistent responding has also been observed on other types of learning probes (Costa, 2009; Costa et al., 2013; Costa et al., 2001; Domeniconi et al., 2007; Schmidt, Franco, Loterio, & Gomes, 2016; Wilkinson & McIlvane, 1997). However, correct responding increases after repeated exposure to exclusion trials (Costa, 2009; Costa et al., 2001; Costa et al., 2013; Domeniconi et al., 2007; Greer & Du, 2015; Wilkinson & McIlvane, 1997). Considering that there is a lack of consensus regarding the number of exclusion trials after which a participant reliably shows learning by exclusion, a second goal of the present study was to explore the effects of the number of exclusion trials (1, 4, 7, or 10) on the learning outcomes of participants who did not reach the learning criterion after the first exclusion trial.

In sum, this study had two purposes: (1) to assess the generality of learning by exclusion to word-object property (adjectives of emotion) relations, and (2) to evaluate the effect on learning outcomes of additive exclusion trials (with reinforcement) after negative outcomes on the first learning outcome tests (in extinction).

Method

Participants

Participants were 21 typically developing preschool children (11 boys and 10 girls, age range, 24-29 months), who were recruited from a public preschool in the state of Sao Paulo, Brazil. The children's listener repertoire fell within their chronological age range, according to their scores on the Peabody Picture Vocabulary Test Revised (PPVT-R, Dunn & Dunn, 1981). Table 1 presents individual participants' gender, age, and language-age equivalent (PPVT-R), and the experimental conditions to which each child was exposed. Six participants were initially exposed to the Noun condition followed by the Adjective condition; eight participants were exposed to the conditions in the reverse order. Another seven participants (gray lines) were exposed to a single condition (two and five to the Noun and Adjective conditions, respectively), due to several factors, including moving to another preschool or city, summer vacations, or refusal to participate. In total, 16 children participated in the Noun condition and 19 in the Adjective condition.

Setting and apparatus

Sessions were conducted individually, 5 days a week, in a quiet office at the children's preschool, and they lasted approximately 10 min. A Macintosh portable computer equipped with MTS software (Dube, 1991) controlled the sequence of the experimental procedure (i.e., the stimulus display on each trial, inter-trial interval (ITI), consequences for correct and incorrect responses, schedule of reinforcement) and recorded responses. Each participant sat in front of the portable computer, and his/her task was, on each trial, to point to one of the three comparison stimuli displayed on the computer screen. After initial instructions, the experimenter sat behind the participant and recorded the child's response using a keyboard. During the procedure, interaction between the child and experimenter was as limited as possible, but when the task was completed, the experimenter closed the computer and played with the child.

Stimuli

Experimental stimuli consisted of spoken Brazilian Portuguese words (auditory samples), and pictures (visual comparisons). Words were recorded by a female voice and presented over the computer's external speakers. Each visual stimulus was presented on a 5 x 5-cm frame in one of the four corners of the computer screen. In the Noun condition, the comparison stimuli were pictures of objects, and the sample stimulus on each trial was the name of one of those objects. In the Adjective condition, the comparison stimuli were pictures of the same puppet (Fafa) with different facial (emotional) expressions, and the sample stimuli were adjectives (the Portuguese equivalents of /angry/, /happy/, /sad/). Auditory samples and visual comparisons were either defined or undefined (novel) stimuli (cf. McIlvane & Stoddard, 1981; Wilkinson & McIlvane, 1997). Defined stimuli were those sample/comparisons that children had already learned to relate pre-experimentally or during baseline training. Undefined stimuli were pseudo-words and pictures of unknown or unfamiliar objects, not previously related, in children's extra-experimental or experimental history. For convenience of expression, stimuli will be indicated by their history (D for defined; U for undefined), followed by the word category (N for noun; A for adjective).

Table 2 shows the stimuli used in each condition. The defined auditory samples were the Portuguese words /bola/, /moto/, and /aviao/ (ball, motorcycle, and airplane) in the Noun condition, and /brava/, /feliz/, and /triste/ (angry, happy, and sad) in the Adjective condition. The defined comparisons were photos of a ball, motorcycle, and airplane in the Noun condition, and photos of Fafa displaying angry, happy, and sad faces in the Adjective condition. The undefined words presented in both conditions were pseudo-words without any meaning in Portuguese: /fapi/, /beva/, /nopa/, and /mupa/ in the Noun condition, and /foba/, /piva/, /mipa/, and /nafu/ in the Adjective condition. The visual undefined stimuli were pictures of objects built for this study and pictures of the puppet Fafa showing a variety of facial expressions for which there are no corresponding words in Portuguese.

Consequences

During the entire experiment, correct responses on training trials were followed by the presentation of colored stars on the computer screen, an ascending chime sound, and the presentation of compliments by the experimenter. A 2-s black screen followed incorrect responses. There was a 0.5-s ITI. Responses on probe trials had no programmed consequences other than the ITI. As an additional consequence (with the goal of maintaining the behavior of coming to the session), after the experimental task, the experimenter played with the child (using drawings, soap bubbles, and other toys) for up to 5 min.

Procedure

The main task was a sequence of MTS trials. On each trial, after the presentation of a spoken word as the sample stimulus, three visual stimuli were simultaneously displayed; the participant's task was to select one of the visual stimuli, conditionally to the spoken word. The operant response was touching the stimulus on the screen. The spoken word was repeated every 2 s until a response occurred. The type of words used as sample stimuli defined two experimental conditions (Noun and Adjective), as shown in Table 2.

Participants were initially exposed to auditory-visual baseline training with three object name-object relations (Noun condition) or three adjective (emotion) names-puppets with facial expressions relations (Adjective condition). Each baseline and its associated tests were presented sequentially, and the order was counterbalanced across participants. For each condition, probes evaluated the performance for two novel relations (name-picture relations in the Noun condition and adjective-picture relations in the Adjective condition). After baseline performances met criterion, exclusion, control, and learning outcome trials for one undefined sample-comparison relation were presented (in extinction, intermixed with reinforced baseline trials). If negative outcomes occurred, additional exclusion trials with reinforcement were presented and learning outcome tests were repeated (maximum of 10 exclusion trials). A second undefined sample-comparison relation was tested using the same procedure. Finally, the baseline and testing procedures were replicated with the second condition (Noun or Adjective).

As illustrated in Fig. 1, the procedure began with pretraining (Phase 1), which was designed to familiarize children with the experimental setting and the auditory-visual task conducted on the computer. Pretraining was followed by three phases of the first condition, and subsequently, by three phases of the second condition (Noun and Adjective conditions were counterbalanced as first or second in this sequence): Baseline Training (Phase 2), Exclusion and Learning Outcome Probes for Relation 1 (Phase 3); and Exclusion and Learning Outcome Probes for Relation 2 (Phase 4). After completing Phase 4 of the first condition, the participant was exposed to the second condition, following the same three phases. As shown in Table 1, the distribution of participants in the Noun and in the Adjective conditions was counterbalanced; therefore, the Noun condition was the first for some participants and the second for other participants, and the same was true for the Adjective condition (but some participants completed only the first condition).

Across the sequence of experimental phases, the reinforcement schedule for correct responding was CRF on all trials of Phase 1 and in all trials of the first four blocks of Phase 2. From the fifth block of Phase 2 to the end of Phase 4, trials were reinforced according to an intermittent reinforcement schedule: random ratio 2 (RR 2). According to this schedule, half of the trials, on average, were reinforced. The position of the three comparison stimuli on the screen varied across trials, thus ensuring that the correct response could occur in all four designated positions.

Phase 1--pretraining Stimuli were pictures of familiar objects and their corresponding names, and were not used in subsequent phases. Each trial began with the simultaneous presentation of the auditory sample and three visual comparisons on the computer screen. The child was instructed to point to the object corresponding to the dictated word. At this point, if necessary, the experimenter would shape the participant's responses, either by physically prompting the correct comparison (i.e., holding the participant's hand and moving it in the direction of the stimulus) or by giving verbal feedback. A block of six trials was programmed. This block was repeated until the participant reached 100% independent correct responding in six consecutive trials.

Phase 2--baseline training Phase 2 was programmed with five types of blocks, each including six to nine trials. Each block was repeated until the participant reached the criterion of 100% correct responses. The sequence of sample stimuli and the position of comparison stimuli were different in each repeated block. The first two blocks had six trials, and the samples were the defined spoken words /bola/ (ball), /moto/ (motorcycle), and /aviao/ (airplane) for the Noun condition, and /brava/ (angry), /feliz/ (happy), and /triste/ (sad) for the Adjective condition. Comparison stimuli were the defined pictures Dl, D2, and D3 for each condition (see Table 2).

The third block introduced the mask, or "blank" comparison (Wilkinson & McIlvane, 1997), among the baseline visual comparison stimuli; that is, a black square covered one of the three comparisons that were simultaneously displayed on the computer screen. This block had nine trials. On the first trial, the black square was small and it gradually increased in size across trials (fading in), covering the entire picture on the ninth trial; it covered the correct comparison (S+) on five trials, and one of the incorrect comparisons (S-) on four trials. According to Wilkinson and McIlvane (1997), the presence of the mask in the comparison array allows the selection of one picture (equivalent to "Yes, this is the picture") or the rejection of all the displayed pictures (equivalent to "No, none of these pictures"). Each sample was presented on three trials in a semi-random order. Comparison stimuli were the three defined visual stimuli on all trials. If errors occurred, the block was repeated from the beginning.

The purpose of the fourth type of baseline-block trials was the stabilization of the auditory-visual matching baseline with the mask as one of the comparisons on all trials. This block had six trials. On each trial, the sample stimulus was a defined spoken word. Two defined pictures and the mask were presented as comparisons. Each sample was dictated twice. The block was repeated until criterion was reached (100% correct responses in two successive presentations of this block). Each session presented a maximum of two blocks; the block was repeated in subsequent sessions until the criterion was met.

The fifth baseline block (six trials) was similar to the previous block, except that the reinforcement schedule changed from CRF to RR 2. The computer program specified the trials in which a consequence for responding was available. On trials with programmed consequences, correct responses produced the stars and the chime sounds for 2 s. Incorrect responses were followed by the ITI and the next trial; the scheduled consequence was canceled without replacement. Given the odd number of trials with programmed consequences (three), a picture was the S+ on approximately half of the reinforced trials, and the mask was the S+ for the other half of trials; this was reversed each time the block was repeated. The block was repeated until criterion (100% correct responses in two consecutive presentations of this block) was reached, with the restriction that no more than three blocks were conducted in the same session.

Phase 3--exclusion and learning outcome probes for relation 1 After participants reached criterion on block 5 of Phase 2, exclusion, control, and learning outcome probe trials were interspersed among baseline trials. Tables 3 and 4 show the stimuli used as the samples and comparisons on each type of probe for the Noun and Adjective conditions, respectively.

The initial Phase 3 probes were conducted in two blocks. The first block had six trials: one exclusion and one control probe (Control Probe 1), interspersed among four baseline (BL) trials. The exclusion probe displayed one undefined spoken word as the sample stimulus (U1); the comparison stimuli were one Undefined Picture (U1), one Defined Picture (D), and the mask (M). The undefined word for the Noun condition was /beva/ (UNI). The undefined word for the Adjective condition was /foba/ (UA1). Selection of the undefined comparison was classified as responding by exclusion (or rejection of the defined and the mask comparisons). However, the selection response could also have occurred under control of the novelty of the undefined visual stimulus. To verily this possibility, the procedure included two types of control trials. Control Probe 1, conducted in Block 1, presented a defined spoken word as the sample stimulus; the comparison stimuli were one defined picture, the same undefined picture presented on the exclusion probe, and the mask. Selection of the undefined stimulus would indicate control by novelty, whereas selection of the defined stimulus would indicate maintenance of baseline, increasing the confidence that control by novelty was not responsible for the participants' response on the previous exclusion probe.

Block 2 of Phase 3 included three learning outcome probes and a second control Probe (Control Probe 2), interspersed among three BL trials. Table 3 shows the stimuli presented on each type of probe trials. The design of learning probes used in the present experiment was based on previous studies (e.g., Costa et al., 2001; Wilkinson & McIlvane, 1997; Wilkinson et al, 1998).

Control Probe 2 verified whether participants would correctly reject defined pictures and select the mask when the dictated word was the same undefined sample previously presented in the exclusion probe (U1). The comparisons were two defined stimuli and the mask. Selecting the mask was an important demonstration that the baseline repertoire was maintained: In the presence of a sample stimulus not related to one of the defined comparisons, the mask effectively controlled the selection response.

For Learning Probe A, the sample was the same undefined word dictated in the exclusion probe (U1), and the comparisons were the corresponding picture (U1), one novel undefined picture (U3), and the mask. If participants had related the two U1 stimuli (theU1 sample and the U1 comparison) on the exclusion trial, they should select U1 once again, and not U3 or the mask, on this probe trial.

Learning Probe B also presented the same undefined spoken word previously presented on the exclusion probe (U1) as the sample stimulus; the comparisons were one novel undefined picture (U5; different from those used on the exclusion probe and on Probe A), one defined picture, and the mask (supposedly covering picture U1). Selecting the mask would be consistent with the possibility that the U1 name-picture relation had been established on the exclusion trial.

On Learning Probe C, the sample was a different undefined word (U3: /nopa/ in the Noun condition and /mipa/ in the Adjective condition; see Table 2); the comparisons were the undefined picture U1, one defined picture, and the mask. Thus, if the participant had learned the relation between U1 name and U1 picture, the alternative consistent with this outcome would be to select the mask. Collectively, responses consistent with learning the U1 name-picture relation on the exclusion trial were the selection of U1 on Probe A, and the selection of the mask on Probes B and C (Table 3).

Additive exclusion trials Rigorously speaking, exclusion responding can only be assessed when the two undefined stimuli are presented for the first time. However, our second objective was to verify the effects of additive sample/ comparison pairings of the two undefined stimuli used on the exclusion trial. If participants did not select the stimulus consistent with learning the U1 (word-picture) relation on all three learning probes after the first exclusion trial, extra blocks were conducted in Phase 3, during which exclusion probes had programmed consequences (that is, responding by exclusion was reinforced; cf. de Rose, de Souza, & Hanna 1996; McIlvane et al., 1992).

Each block had nine trials and presented three exclusion (reinforced) trials, interspersed among six BL trials. The arrangement of comparison stimuli on these trials was the same as the one used in the original exclusion trial: the same undefined comparison (U1), one defined stimulus (in this case, each one of the three trials presented a different stimulus as S-: Dl, D2, or D3), and the mask. The block of reinforced exclusion trials was followed by a block of seven trials, which assessed Control Probe 2 and learning outcome Probes A, B, and C, interspersed among three BL trials.

If the learning criterion was not reached, these two blocks of additive exclusion trials and probe trials were repeated once or twice. Thus, for each participant, the number of exclusion trials varied (from 1 to 10), depending on whether they reached the learning criterion. The maximum was 10 exclusion trials, but not all participants reached criterion on the learning outcome probe, even after 10 trials. Thus, Phase 3 ended after the participant reached the criterion of 100% correct responses on all learning probes or had been exposed to three extra blocks of reinforced exclusion trials.

Phase 4--exclusion and learning outcome probes for relation 2 Phase 4 followed the procedure used in Phase 3, except that the focus was a second word-picture relation, using additional undefined stimuli. On exclusion probes, the sample (U2) was the word /fapi/ in the Noun condition and the word /piva/ in the Adjective condition; the corresponding picture was the undefined picture U2 of each condition (see Table 3). On learning probes, the spoken undefined words were /fapi/ and /mupa/ in the Noun condition, and /piva/ and /nafii/ in the Adjective condition. The undefined comparison stimuli in Phase 4 probes were the pictures U2, U4, and U6.

Again, if participants did not respond correctly on all three learning probes after the first exclusion trial in Phase 4, they were exposed to up to three extra blocks, each containing three reinforced exclusion trials interspersed with six BL trials. Learning outcome probes were conducted after each extra block. This phase was completed after participants reached the criterion of 100% correct responses on all learning probes, or after they had been exposed to three extra blocks of exclusion teaching trials (a maximum of 10 exclusion trials).

Intra-subject replication After participants completed all three phases of the first condition (Noun or Adjective), they were exposed to the same sequence of procedures (Phases 2, 3, and 4) for the second condition (Adjective or Noun). For various reasons, seven of the 21 participants were exposed only to the first condition (see Table 1, and the Results section).

Results

A stable baseline is required for conducting exclusion probes. The procedure effectively established the two auditory-visual baselines of three object name-object relations (Noun condition) and three adjective (emotion) names-puppets with facial expressions (Adjective condition). Nevertheless, as can be seen in Fig. 2, more trials to criterion were required to establish the Adjective baseline (black bars) than the Noun baseline (white bars), considering the intra-subject replications (individual data of participants exposed to both conditions), intersubject replications (data of different participants under each condition), and the average data (two rightmost bars in Fig. 2). In the Noun condition, most children reached criterion with minimum exposure to each block, independent of order (average: 1.5 exposures), whereas in the Adjective condition, several participants required more exposures (average: 3.3 exposures).

The trend toward more training for adjectives occurred especially in Blocks 1 and 4, but the number of block repetitions was also high in other instances, even when the Adjective condition came second (e.g., P6 in Block 2; P4 and P6 in Block 4; P3 in Block 5). A two-sample t test showed a significant difference between the average number of blocks in the Noun and Adjective conditions (p = 0.016). Furthermore, two-sample / tests also revealed a significant difference between the Noun and Adjective conditions for each block (p < 0.05 for Blocks 3 and 5; p = 0.05 for Block 2; p < 0.005 for Blocks 1 and 4). In summary, all children reached criterion in baseline, and thus advanced to the exclusion and learning probes.

With regard to the first question of this study, Table 5 shows that the occurrence of responses consistent with exclusion was high on the first exclusion probe: 61 responses (87.1%) in 70 opportunities. More specifically, participants selected the undefined stimulus on 94% of the opportunities in the Noun condition (30/32), whereas they selected the undefined stimulus on 82% of the opportunities (31/38) in the Adjective condition. Therefore, participants responded by exclusion in both the Noun and the Adjective conditions.

On the Noun exclusion probes, the two participants who did not respond under the control of the reject relation selected the mask (one participant when the novel sample was beva, and the other when it was fapi). On the Adjective exclusion probes, three participants did not select the novel face foba (two chose the mask and one chose the defined stimulus), and four participants did not select the novel face piva (two selected the defined stimulus and two selected the mask; see Table 5).

For economy of expression, the term "correct responses" will be used to indicate selection responses consistent with baseline training and with the emergence of a novel relation between the undefined sample and comparison, presented for the first time on the exclusion trial.

Figure 3 shows data for individual participants on both the exclusion and control trials. The exclusion data (filled squares) refer to all exclusion trials (reinforced and non-reinforced), showing that almost all of the participants chose the undefined stimulus for both nouns and adjectives. The lowest value (average proportion of correct responses for the two relations in each condition) was 0.75. In the Noun condition, only 3 of 16 participants (P2, P4, and P8) had a score lower than 1.0 in the proportion of selections of the undefined stimuli (0.75, 0.80, and 0.93, respectively). In the Adjective condition, 12 of 19 participants selected the undefined stimuli on all exclusion trials. The remaining seven participants showed high proportions of exclusion responding: 0.95 for P2, P13, P14, and P19; 0.90 for P4 and P10; and 0.75 for P17.

Data from the control probes, which aimed to increase the confidence that responses on the first trial were truly controlled by exclusion (reject relations), showed that, on Control Probe 1 (open circles), all participants selected the baseline (defined) stimulus when the dictated sample was a defined word and the comparisons were one defined stimulus, one undefined stimulus, and the mask. Additionally, on Control Probe 2 (open triangles), participants tended to select the mask when an undefined word was dictated, but the array of comparisons displayed only defined stimuli with the mask. Control Probe 2 was presented four times at the most (after 1, 4, 7, or 10 exclusion trials) for each relation. Almost all participants selected the mask on every presentation of this type of trial in the Noun condition; the only exceptions were P10, PI 1, and PI4, who showed the following proportions of correct responses (choosing the mask): 0.75, 0.86, and 0.75, respectively. In the Adjective condition, only three (of 19) participants did not choose the mask on all opportunities: P3 and P9 had a proportion of 0.88 correct, and P14 reached 0.25.

The learning outcome probes assessed whether responding by exclusion would engender learning of novel relations between the undefined sample and the undefined comparison presented on those trials. As detailed in the Procedure section (see Tables 3 and 4), three types of learning outcome probes were implemented (each type of probe was conducted with different stimuli--therefore, there were several probes of each type, but for convenience, from now on each type will be referred to as Probe A, Probe B, or Probe C).

Figure 4 shows the average proportion of correct responses and the standard error of the mean for all participants on learning outcome Probes A, B, and C in the Noun (upper panel) and Adjective (bottom panel) conditions. Each panel presents the average proportion of correct responses for two word-picture pairs (beva and fapi on top; foba and piva, at the bottom). For each probe, the white bar represents the measure of learning after the first exclusion probe, and the black bar represents results after all exclusion trials (both non-reinforced and reinforced).

On Probe A, the sample was the same undefined word dictated on exclusion trials, but the comparison array displayed two undefined pictures (the same undefined picture presented on exclusion trials and a second undefined picture), and the mask (see Tables 3 and 4). After the first exclusion trial, the mean proportion of correct responses was 0.56 for Nouns and 0.63 for Adjectives. Proportions of correct responses increased for both the Noun and Adjective conditions with the additive reinforced trials (0.68 and 0.71, respectively). In the instances of inconsistent responding, participants selected the most novel undefined picture.

Probe B also displayed the same undefined word used as the sample on the exclusion trial, but the corresponding picture was not among the comparisons (See Tables 3 and 4). A different undefined comparison was presented with a defined stimulus and the mask (S+). The mean proportion of correct responses (selecting the mask) after the first trial was 0.12 in the Noun condition and 0.32 in the Adjective condition. Proportions increased only marginally after all exclusion trials for both Nouns (to 0.15) and Adjectives (to 0.34). Participants tended to select the undefined comparison stimulus and not the mask.

Finally, the proportion of correct responding on Probe C after the first exclusion trial was 0.22 for Nouns and 0.29 for Adjectives. Additive reinforced trials resulted in an increase in the proportion of correct responses in the Noun condition (to 0.35) but not in the Adjective condition (to 0.28). On this probe, the sample was a novel undefined word, and comparisons were the undefined picture corresponding to the sample word dictated on the exclusion trials, one baseline (defined) stimulus, and the mask (S+). Participants mostly chose the undefined picture instead of the mask.

An ANOVA was conducted to test possible effects of the type of learning outcome probe in each condition. A significant main effect was found for the Noun condition, F(2, 30) = 25.32, p < 0.001. Pairwise comparisons were conducted using Bonferroni's correction. There was a significant difference in the average proportion of correct responses between Probes A and B (p < 0.001), A and C(p = 0.005), and B and C (p < 0.05). A main effect of learning outcome probe was also found in the Adjective condition, F(2, 36)= 19.01, p<0.001. Pairwise comparisons indicated that the average proportion of correct responses on Probe A was significantly different from Probe B (p = 0.001) and Probe C (p < 0.001). No significant difference was found between Probes B and C. A Mann-Whitney test did not show significant differences in the average proportion of correct responses between the Noun and the Adjective conditions.

Table 6 shows the percentage of correct responses for each word-picture relation as a function of the number of exclusion trials (1,4,7, or 10). The number of participants after each one of these parameters decreased because participants who reached the learning criterion (correct responses for all three learning probes) were not exposed to extra blocks of additive exclusion trials.

In general, the proportion of correct responding during Probe A was equally high for both conditions from the beginning: The average proportion for the two Noun relations on all trials was 70.0% (range, 44-94%), and the average percentage for the two Adjective relations trials was 70.7% (range 42-92%). The two lowest percentages occurred on the first probe trial (42% for the Adjective relation foba and 44% for the Noun relation fapi).

After the first exclusion trial, the percentage of correct responses varied on Probe A across relations from 42% (Foba) to 84% (Piva). On the second Probe A, after four exclusion trials, percentages varied from 63% (Foba) to 94% (Fapi). On the third Probe A, after seven trials, percentages varied from 63% (Fapi) to 92% (Piva). Finally, on the fourth Probe A, after 10 trials, percentages varied from 67% (Fapi) to 85% (Beva). There was significant variability in the percentage of correct responses across trials, but, in general, the percentage tended to increase as a function of the number of exclusion trials.

For Learning Probes B and C, percentages of correct selections were low. On Probe B, the percentages for the two Noun relations varied from 6-33%, and for Adjective relations the percentages varied from 24-50%. Probe C resulted in percentages varying from 15-60% for Nouns, and from 0-35% for Adjective relations. As can be seen in Fig. 4, the main effect of the additive exclusion trials was an increase in the percentage of correct responses on Probe A for both Noun and Adjective conditions.

Table 7 summarizes the results regarding learning outcome probes for individual participants: "No" indicates that the learning criterion was not reached. A number (1, 4, 7, or 10, see "Procedure") indicates the number of exclusion trials before the participant reached the learning criterion. Outcomes were evaluated as positive or negative by considering all three types of learning probes together (All) or considering only the results of Probe A.

Considering all three-trial types, participant P10 showed positive outcomes for all four relations (two noun-picture relations and two adjective-picture relations), whereas P20 showed negative outcomes for all four relations. The remaining participants showed positive outcomes for some, but not all, relations. For the Noun condition, four participants showed positive outcomes for the beva relation, after a varying number of exclusion trials: one (one participant), four (two participants), or ten (one participant). Four participants showed positive outcomes for the fapi relation after four (one participant), seven (two participants), or ten exclusion trials (one participant). In the Adjective condition, five participants reached learning criteria for the foba relation, after four exclusion trials (two participants) or after ten trials (three participants). Nine participants showed evidence of learning the piva relation, after one (four participants), four (three participants), or ten exclusion probes (two participants).

Given the low proportion of correct responses on Probes B and C, learning outcomes were reassessed considering only correct responses on Probe A (second column for each relation in Table 7). This analysis yielded a different scenario: 15 and 11 participants (of 16) showed positive outcomes for the name-picture relations beva and fapi, respectively, and 15 participants (of 19) showed positive outcomes for the adjective-picture relations foba and piva. In sum, 14 of 21 participants were exposed to the two conditions (Noun and Adjective), and 12 showed positive results on Probe A for at least one relation of each condition. One participant (P4) showed correct responses for only Adjective relations, and another (P10) showed correct responses for all four relations. The two participants who were exposed only to the Noun condition (P7 and P8) showed positive results for the two novel relations (beva and fapi). Among the five participants who were exposed only to the Adjective condition, two (PI7 and P21) were successful on Probe A for the two novel relations (foba and piva), and two were successful on the probes for one of the two relations (P18 and P19). Participant 20 was the only one who showed no positive results on Probe A after ten exclusion trials with each of the two relations.

Discussion

This study aimed (1) to assess the generality of responding and learning by exclusion to word-object property (adjectives of emotion) relations, and (2) to evaluate the effect of additive exclusion trials (with reinforcement) on learning outcomes after negative results on the first outcome probes (in extinction).

The Adjective condition used in this study addressed the first question; the Noun condition was used as a control. Considering the generality of existing data on exclusion responding after a baseline of noun-object relations, replication of the typical exclusion result in participants exposed to both the Noun and Adjective conditions would attest to the appropriateness of a procedure that includes both types of relations. The results, obtained with 24- to 29-month-old children, indicated that more trials to criterion were required to establish the Adjective baseline, but that exclusion and learning outcome results from the Noun and Adjective conditions were comparable, suggesting that the phenomenon of learning by exclusion has generality across stimulus (lexical) classes.

The results from the Noun condition replicated previous findings on word-object relations, suggesting the generality of data regarding exclusion responding with participants younger than 36 months of age (Costa, 2009; Costa et al., 2001; Dixon, 1977; Domeniconi et al., 2007; Ferrari et al., 1993; McIlvane & Stoddard, 1981; Schmidt et al., 2016; Wilkinson & McIlvane, 1997; Wilkinson, Rosenquist, & McIlvane, 2009). The extensive replication with participants in this study not only increases the generality of data on emergent responding in a simulated route for the acquisition of novel word-object relations, but also confirms the validity of the procedure for evaluating this pattern of responding in a situation that simulates the acquisition of word-object property relations.

Results from the Adjective condition were consistent with those in the Noun condition, extending previous findings to this other type of word-environment relation (i.e., word-property). Moreover, our data are consistent with results from studies using other stimulus properties, such as color, shape, and texture, and their respective names (Carey & Bartlett 1978; Costa, 2009; Costa et al., 2010; Heibeck & Markman, 1987; Holland et al, 2015; Rice, Buhr, & Nemeth, 1990).

Despite the predominant selection of the novel comparison on the first exclusion trial, during which an undefined noun or adjective was dictated (94% and 82%, respectively), the few cases of inconsistent responding occurred more frequently in the Adjective than in the Noun condition. These inconsistencies, mainly selection of the mask, may have been due to uncontrolled variables such as deleterious effects of errors (McIlvane & Dube, 2003; Mueller, Palkovic, & Maynard 2007; Sidman, 1985; Stoddard, de Rose, & McIlvane, 1986; Stoddard & Sidman, 1967), which were more frequent during baseline training in the Adjective condition. When the comparison corresponding to the sample was not displayed in the comparison array, the stimulus control topography consistent with the defined reinforcement contingencies was selecting the mask (as a S+). However, on baseline trials involving incorrect selection of the mask, children's responses may have been under the control of irrelevant stimulus control topographies (e.g., selecting the mask after a previously reinforced selection of this stimulus, or selecting the mask if it was displayed in the same position as the S+ on a previous trial). Later, the presence of novel stimuli on exclusion trials or on learning outcome trials may have set the occasion for the occurrence of potentially conflicting stimulus control topographies.

The issue of stimulus control topographies that may develop during training and on probe trials is pertinent to our second research question, which addresses a long-standing conceptual and related methodological issue (cf. Wilkinson & McIlvane, 1997; Wilkinson et al., 1998). Sample/S+ control could establish a direct relation between the two undefined stimuli (the sample and comparison), whereas sample/S- control would not favor the establishment of this relation (cf. Johnson & Sidman, 1993). As argued by McIlvane et al. (2000), both sample/S+ and sample/S- control are necessary for the establishment of a relation between a specific sample stimulus and the comparison assigned as correct by the reinforcement contingencies and for the emergence of derived relations among those stimuli.

More specifically, most children demonstrated exclusion responding, as shown by selecting the undefined comparison stimulus in the presence of the undefined auditory sample (not responding to the defined stimulus or the mask comparison, and responding appropriately in two control tests). On the other hand, the majority of children did not respond consistently on all learning outcome tests, which were collectively designed to assess whether participants learned specific word-object and word-object property relations (i.e., conditional discrimination with both sample/S+, and sample/S- relational control).

The additive exclusion trials and the three different learning outcome probes addressed our second question. Probe A presented the undefined sample and comparison (both presented on the exclusion trial), a new undefined comparison, and the mask. Selection of the undefined comparison displayed on the exclusion trial would be consistent with having learned to relate a specific name to a specific object/object property. Probe B again presented the undefined word from the exclusion trial as the sample, and one defined stimulus, one novel undefined stimulus, and the mask as comparisons. Selections of the mask in this case would be consistent with having learned to relate a specific name to a specific object/object property. Probe C presented one new undefined stimulus as sample, one defined comparison, the undefined comparison presented on the exclusion trial, and the mask. Selection of the mask would be consistent with having learned that a name previously related to another stimulus, which is not available in the array of comparisons, cannot be related to a second object/object property. After the first exclusion trial (non-reinforced), the majority of children showed positive outcomes on Probe A and negative outcomes on Probes B and C (Table 6). These results replicate the findings of previous studies (Costa, 2009; Costa et al., 2001; Domeniconi et al., 2007; Wilkinson & McIlvane, 1997; see also Schmidt et al., 2016, Probes B and C).

The additive exclusion trials (with reinforcement) increased the percentage of positive learning outcomes on Probe A, from an average of 60% (average for all four word-picture relations) after the first trial to 70.4% after all reinforced trials. Considering all probes together (A, B, C), the mean percentage increased from 36.2% to 41.4% (calculations based on all opportunities presented on Table 6). These results suggest that additional exclusion trials with reinforcement influenced learning outcomes; however, negative outcomes were still more probable than positive outcomes when the criterion for learning included all probe trial types. Given the high occurrence of negative outcomes, most previous studies have concluded that a single exclusion trial is not sufficient to establish a specific name-object relation. The same conclusion would hold for the larger number of (additive) exclusion trials: despite the overall increase in the percentages of responding consistent with learning, results of Probes B and C remained negative.

Similar contradictory results on learning outcome probes have been found in previous studies conducted with participants of different age groups (Costa, 2009; Costa et al., 2010, 2013; Domeniconi et al. 2007; Schmidt et al, 2016; Wilkinson & McIlvane, 1997). For comparison purposes, Table 8 (adapted from Costa, Domeniconi, & de Souza, 2014) summarizes our results on learning outcome Probes A, B, and C, along with results from seven other studies. These studies were selected because they used the mask as one of the comparisons, and we had access to the original data (via published papers or because the dataset was collected in our lab). Percentages of correct responding on Probe A were similarly high across studies (61% or higher). Results from Probes B and C varied considerably across studies, but they were mostly negative (except for two studies, i.e., Costa et al., 2013, Probe B; Domeniconi et al., 2007; Probe C).

Probes B and C assessed the effects of replacing one of the stimuli potentially related by exclusion: Probe B replaced the undefined comparison from the exclusion trial by a different undefined stimulus, and Probe C replaced the sample word by a different undefined dictated word. If the stimulus-stimulus relation was well established in the exclusion trials, it should not be disturbed by the introduction of a different undefined stimulus (as one of the comparisons on Probe B, or as the sample on Probe C). In both cases, selecting the mask would indicate that the participants learned to relate a specific name to a specific object/adjective by exclusion; however, participants tended to select the novel stimulus in both probes. The contradictory results from Probes A, B, and C raise questions regarding the appropriate experimental arrangement for determining learning outcomes of exclusion responding. In other words, what are the necessary and sufficient tests to determine learning outcomes of exclusion responding, including the emergence of a conditional discrimination?

Clearly, in the present study, the selection of the undefined stimuli on Probes B and C could be interpreted as control by novelty, but this interpretation is incompatible with results from other probes: participants did not select the undefined stimulus presented for the first time on Probe A or the undefined stimulus presented on Control Probe 1. In order to understand these response patterns, it seems important to consider not only the specific comparison stimulus selected by a participant, but also the specific arrangement of stimuli on each probe type (the context for responding and the potential controlling relations involved in each arrangement), as well as the experimental history. It is possible that the blank comparison procedure and reinforced exclusion trials used in the current study biased responding away from the mask, either when an undefined stimulus was presented as a sample or when an undefined comparison was available (stimulus arrangements). The only feedback given for responding in the presence of an undefined sample occurred on reinforced exclusion trials when a non-mask stimulus was selected (experimental history). Participants indeed selected the mask on Control Probe 2, during which the sample was an undefined stimulus, but in this case there was no undefined comparison available (comparisons were two defined comparisons and the mask).

These findings are also suggestive of temporal discriminations in which all undefined stimuli may become equivalent due to their late introduction in the experimental sequence. If that were the case, any undefined comparison in the presence of an undefined sample would control the selection response. The formation of such a class of equivalent stimuli could interact with a potential bias away from the mask under particular arrangements involving undefined sample and comparisons. Future studies should consider experimental arrangements that might minimize or avoid this potential confound. For example, undefined comparisons could be presented as distractors during baseline (as done by Dixon, 1977), and the undefined samples would be saved for probes. Additionally, the arrangement used as Control Probe 2 in the present study could be used as a teaching tool: reinforcing the selection of the mask in the presence of different undefined samples (which would not be used on probes) might strengthen responding to the mask. In both cases, the repertoire required for probes would have been established during baseline training; the probes would only introduce novel stimuli.

One limitation of the present study is that we did not use discrimination or naming probes, which would be interesting alternatives in future studies on exclusion with different lexical classes. A discrimination probe contrasts undefined stimuli, previously introduced on exclusion probes, as samples and comparisons (cf. Wilkinson et al., 1998). This arrangement allows for the evaluation of response accuracy in the absence of the defined stimulus. Positive results on discrimination probes could be interpreted as suggesting the emergence of novel conditional discriminations; that is, participants select a specific comparison under the control of the corresponding sample, as a result of having responded to those same stimuli on the exclusion trials. For example, if three novel relations are introduced on exclusion probes, the corresponding discrimination probes would present one undefined sample (1, 2, or 3), and all three undefined comparisons (1,2, and 3) simultaneously. This type of probe has produced positive results (e.g., de Rose et al., 1996; Ferrari et al., 1993; McIlvane & Stoddard, 1981). In these studies, discrimination probes were conducted after an extensive history of reinforced exclusion trials, designed as a teaching procedure. For this reason, responding by exclusion was followed by potentially reinforcing consequences (exclusion procedure), which may have generated the positive results. In future studies, discrimination probes could also be used after a single non-reinforced exclusion trial. Additionally, discrimination probes should be systematically included in a standard protocol for the study of responding and learning by exclusion.

Considering the potential of exclusion learning for vocabulary acquisition, and given the use of spoken words as sample stimuli in baselines of conditional discriminations, one important question is whether the child, who has selected undefined pictures conditionally to spoken words, would name (tact) the picture. Several studies have reliably demonstrated the emergence of a speaker's repertoire; that is, participants learned to say the name of (to tact) pictures (Carey & Bartlett 1978; Costa 2009; Costa et al., 2013; Ferrari et al., 1993; Almeida-Verdu et al. 2008), foods (McIlvane et al., 1984; McIlvane & Stoddard, 1981), and printed words (textual behavior; e.g., de Rose et al., 1996) previously presented as comparison stimuli on exclusion trials. These positive results can be interpreted as the emergence of novel discriminated verbal operants (Skinner, 1957) derived from conditional discriminations established by exclusion. For this reason, naming probes could be one relevant additional tool to evaluate learning outcomes of exclusion responding.

In addition to general conclusions about the two questions the study sought to answer, further aspects of the present research are relevant for discussion. First, the assessment of exclusion responding requires a stable baseline to provide defined stimuli, which can be contrasted with the undefined stimulus on exclusion probes (Dixon, 1977; McIlvane & Stoddard, 1981). Most previous studies on exclusion responding have used familiar spoken words as sample stimuli in a baseline of auditory-visual conditional discriminations, in order to guarantee that participants had already related the words to the corresponding visual stimuli. Thus, the aim of baseline training is to strengthen these relations and to ensure that participants have the appropriate repertoire regarding the experimental task. In this study, baseline training in the Noun condition yielded a good illustration of this procedure and of typical results: participants met criterion in the first or second block of training, demonstrating that they had already related the words to the corresponding objects. Baseline training in the Adjective condition used the same procedure as the one used in the Noun condition, except that sample words were related to different facial expressions of a puppet. Nonetheless, most participants who were successful during Noun baseline training did not consistently relate the adjectives to the corresponding faces at the beginning of training: they required longer training to reach criterion in this condition, as shown in Figure 2. The question that may arise is whether these results were due to an inappropriate choice of those particular stimuli (either the words, the pictures, or both), or whether they reflect a more general aspect of language development, as suggested by studies showing differential acquisition across time of lexical classes (Gentner, 1982; Golinkoff et al., 1994; Poulin-Dubois & Graham, 2007).

A close examination of the visual stimuli used in both conditions suggests a parsimonious account of these findings. As shown in Table 2, stimuli in the Noun condition differed in all respects (color, shape, orientation in space), whereas stimuli in the Adjective condition had many overlapping components with only minor differences in facial expressions across stimuli (e.g., mouth shapes, eyebrow shapes, eyes opened or closed). Differences in the complexity of visual stimuli likely involve different levels of discriminability. Therefore, there is a strong possibility that the baseline acquisition difference between the Noun and Adjective conditions was related to the discriminability of the comparisons in those conditions (Carter & Eckerman, 1975; Sema, Foran, & Tran, 2016). In such case, discrimination difficulty, rather than lexical class per se, could help to explain this finding.

Another limitation of this study was the absence of a pretest to evaluate the conditional discriminations used as baseline. It is possible that in the test, our participants would have shown better performance in conditional discriminations involving nouns than adjectives (that is, they would have shown that word-object relations had been previously learned, but word-facial expression relations had not). Such potential results could have suggested the choice of other stimuli (words and/or pictures) for the Adjective condition; alternatively, baseline data could demonstrate more clearly why participants needed more trials to achieve baseline criterion in the Adjective condition than in the Noun Condition.

One possible future direction, consistent with the interpretation based on the discriminability of visual stimuli, would be to use stimuli involving more discriminable differences. Another possibility would be to assess children's word recognition repertoire, and use only words that can be recognized as familiar by participants. A better strategy for assessing and comparing responding and learning by exclusion under baselines involving different lexical classes would perhaps be a combination of both alternatives: using familiar and easy-to-discriminate stimuli, thus ensuring that all baselines would have a similar experimental history.

An alternative, which could prevent potential effects of previous experimental history, would be the adoption of pseudo-words and unfamiliar visual stimuli during baseline training. This strategy would allow the assessment of the amount of training necessary to establish baseline relations and to verify any potential differences in the process of acquiring conditional discriminations. On one hand, a similar outcome in the acquisition of baseline relations in both conditions would allow a better comparison of exclusion probes results. On the other hand, if results replicate a difference between Noun and Adjective baselines, this would suggest that the difference was not due to the familiarity of specific words or the discriminability of the comparisons, but to the type of relation to be learned (i.e., word-object or word-property relations). Both outcomes would contribute to a better understanding of developmental and learning routes for the acquisition of different lexical classes.

One positive aspect of the method chosen for the present study, which is relevant for the evaluation of exclusion responding in young children, was the use of participants as their own control in the Noun and Adjective conditions, which allowed for intra-subject comparisons of the two word functions. The order of conditions was counterbalanced between subjects, and no significant differences were found between the two orders. Although one-third of the participants had left the experiment before completing the second condition, their results confirmed those of other participants under that particular condition (inter-subject replication). This finding provides strong evidence that responding by exclusion is a broad phenomenon, not restricted to name-object relations (Carey & Bartlett 1978; Costa, 2009; Domeniconi et al., 2007; Holland et al., 2015; Wilkinson et al., 2000).

The use of control probes, developed in previous studies (McIlvane & Stoddard, 1981; Wilkinson & McIlvane, 1997; Wilkinson et al., 1998), is relevant as a complement to exclusion probes; they can support the identification of controlling variables and provide cues for the interpretation of results from learning outcome probes. For example, the high percentage of correct responses on Control Probe 1 suggests that the selection of the undefined comparison on exclusion probes was effectively under the control of the defined stimuli (reject relation), and not under control of novelty. On that probe, comparison stimuli were the same as those presented on the exclusion probe, but the sample was one defined word. In this case, participants systematically selected the corresponding defined stimulus. If novelty alone played a role, at least, some participants would have selected the undefined stimulus on that probe. On Control Probe 2, participants also showed a high percentage of "correct" responses: they systematically selected the mask when the sample was the same undefined spoken word presented on the previous exclusion trial, and comparisons were two defined stimuli and the mask. The reliable selection of the mask suggests that participants had effectively learned the function of this stimulus (S+ or S-, depending on the sample stimulus), and were able to select it in the presence of an undefined sample (spoken word). Additionally, the accurate performance on this probe type increases the confidence that the selection of the undefined stimulus on exclusion probes was not a by-product of a lack of repertoire to respond to the mask.

In future studies, the combined results of different types of probes, which can potentially replicate each other and reduce inter-subject variability, would be valuable for isolating the various dimensions of responding and learning by exclusion.

DOI: 10.1007/s40732-016-0213-0

[mail] Thais Arantes Ribeiro

thais.arantesr@gmail.com

Compliance with Ethical Standards

Funding This research was part of the scientific program of the National Institute of Science and Technology on Behavior, Cognition, and Teaching (INCT-ECCE), supported by the National Council for Scientific and Technological Development (CNPq; Grant #573972/ 2008-7), and by the Sao Paulo Research Foundation (FAPESP; Grant #08/57705-8). T. A. Ribeiro was supported by a Master's fellowship from FAPESP (Grant #2010/13911-3). T. P. Gallano was supported by a scholarship for undergraduate research from CNPq (Grant #147482/2011-9). Deisy de Souza was supported by a Research Productivity grant from CNPq. We thank Aline Roberta da Costa and William McIlvane for their support of this research. We also thank Cammarie Johnson (at New England Center for Children) for comments and suggestions that greatly improved the manuscript.

Conflict of Interest The authors declare that they have no conflict of interest.

Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent Informed consent was obtained from individual participants included in the study.

References

Almeida-Verdu, A. C., Huziwara, E. M., Souza, D. G., Rose, J. C., Bevilacqua, M. C., Lopes, J., Alves, C. O., & McIlvane, W. J. (2008). Relational learning in children with deafness and cochlear implants. Journal of the Experimental Analysis of Behavior, 89, 407-424.

Beran, M. J., & Washbum, D. A. (2002). Use of exclusion by a chimpanzee (Pan troglodytes) during speech perception and auditory-visual matching to sample. Behavioral Processes, 83, 287-291.

Carey, S., & Bartlett, E. (1978). Acquiring a single new word. Papers and Reports on Child Language Development, 15, 17-29.

Carr, D. (2003). Effects of exemplar training in exclusion responding on auditory-visual discrimination tasks with children with autism. Journal of Applied Behavior Analysis, 36, 507-524. doi: 10.1901 /jaba.2003.36-507.

Carter, D. E., & Eckerman, D. (1975). Symbolic matching by pigeons: rate of learning complex discriminations predicted from simple discriminations. Science, 187, 662-664.

Choi, S. (2000). Caregiver input in English and Korean: use of nouns and verbs in book-reading and toy-play contexts. Journal of Child Language, 27, 69-96.

Cipolla, N. S., Domeniconi, C., & Machado, A. (2014). Responding by exclusion in temporal discrimination tasks. Journal of the Experimental Analysis of Behavior, 101, 215-229.

Clark, E. V. (1983). Meanings and concepts. In P. H. Mussen (Series Ed.) & J. H. Flavell & E. M. Markman (Vol. Eds.), Handbook of Child Psychology: Vol. 3. Cognitive Development (pp. 787-840). New York, NY: Wiley.

Clark, E. Y. (1987). The principle of contrast: A constraint on language acquisition. In B. MacWhinney (Ed.), Mechanisms of Language Acquisition (pp. 1-34). Hillsdale: Erlbaum.

Clement, T. S., & Zentall, T. R. (2003). Choice based on exclusion in pigeons. Psychonomic Bulletin & Review, 10. 959-964.

Costa, A. R.A. (2009). Aprendizagem de relagoes condicionais auditivo-visuais por exclusao com crianqas de 24 a 36 meses [Learning by exclusion of auditory-visual conditional relations in children aged 24- to 36-months], (Research Report FAPESP No. 07/53575-0). Sao Carlos, SP.

Costa, A. R. A., de Rose, J. C., & de Souza, D. G. (2010). Interferencia de variaveis de contexto em sondas de exclusao com substantivos e verbos novos [Interference of contextual variables on exclusion probes of novel names and verbs]. Acta Comportamentalia, 18, 35-54.

Costa, A. R. A., Domeniconi, C., & de Souza, D. G. (2014). Controle de estimulos, mapeamento simbolico emergente e aquisi?ao de vocabulario [Stimulus control, symbolic emergent mapping, and vocabulary acquisition]. In E. J. C. de Rose, M. S. C. A. Gil, & D. G. de Souza (Eds.), Comportamento simbolico: bases conceituais e empiricas (pp. 269-308). Marilia: Oficina Universitaria / Sao Paulo: Culture Academica.

Costa, A. R. A., Grisante, P. C., Domeniconi, C., de Rose, J. C., & de Souza, D. G. (2013). Naming new stimuli after selection by exclusion. Paideia, 23, 217-224.

Costa, A. R. A., Wilkinson, K. M., McIlvane, W. J., & de Souza, D. G. (2001). Emergent word-object mapping by children: further studies using the blank comparison technique. The Psychological Record, 5, 343-355.

de Rose, J. C., de Souza, D. G., & Hanna, E. S. (1996). Teaching reading and spelling: Exclusion and stimulus equivalence. Journal of Applied Behavior Analysis, 29, 451-469.

de Souza, F. M., & Schmidt, A. (2014). Responding by exclusion in Wistar rats in a simultaneous visual discrimination task. Journal of the Experimental Analysis of Behavior, 102, 346-352.

Diesendruck, G. (2005). The principles of conventionality and contrast in word learning: an empirical examination. Developmental Psychology, 41, 451-463.

Dixon, L. S. (1977). The nature of control by spoken words over visual stimulus selection. Journal of the Experimental Analysis of Behavior, 21, 433-442.

Dixon, M. H., Dixon, L. S., & Spradlin, J. E. (1983). Analysis of individual differences of stimulus control among developmentally disabled children. In K. D. Gadow & I. Bialer (Eds.), Advances in Learning and Behavioral Disabilities (pp. 85-110). New York: JAI Press.

Domeniconi, C., Costa, A. R. A., de Souza, D. G., & de Rose, J. C. (2007). Responder por exclusao em cnangas de 2 a 3 anos em uma situafao de brincadeira [Exclusion responding by 2 to 3 year-old children in a play setting]. Psicologia: Reflexao e Critica, 20, 342-350.

Dube, W. V. (1991). Computer software for stimulus control research with Macintosh computers. Experimental Analysis of Human Behavior Bulletin, 9, 28-30.

Dunn, L. M., & Dunn, L. M. (1981). Peabody picture vocabulary test--revised. Circle Pines: American Guidance Service.

Ferrari, C., de Rose, J. C., & McIlvane, W. J. (1993). Exclusion vs. selection training of auditory-visual conditional relations. Journal of Experimental Child Psychology, 56, 49-63.

Gentner, D. (1982). Why nouns are learned before verbs: Linguistic relativity versus natural partitioning. In S. A. Kuczaj (Ed.), Language development: Vol. 2. Language, thought, and culture (pp. 301-334). Hillsdale: Erlbaum.

Golinkoff, R. M., Hirsh-Pasek, K., Bailey, L. M., & Wenger, N. R. (1992). Young children and adults use lexical principles to learn new nouns. Developmental Psychology, 28, 99-108.

Golinkoff, R. M., Mervis, C. B., & Hirsh-Pasek, K. (1994). Early object labels: the case for a developmental lexical principles framework. Journal of Child Language, 21, 125-156.

Greer, R. D., & Du, L. (2015). Experience and the onset of the capability to learn names incidentally by exclusion. The Psychological Record, 65, 355-373.

Hall, D. G., & Graham, S. A. (1999). Lexical form class information guides word-to-object mapping in pre- schoolers. Child Development, 70, 78-91.

Heibeck, T. H., & Markman, E. M. (1987). Word teaming in children: An examination of fast mapping. Child Development, 58, 1021-1034.

Herman, L. M., Richards. D. G., & Wolz, J. P. (1984). Comprehension of sentences by bottlenosed dolphins. Cognition, 16, 129-219.

Holland, A., Simpson, A., & Riggs, K. J. (2015). Young children retain fast mapped object labels better than shape, color, and texture words. Journal of Experimental Child Psychology, 134, 1-11.

Johnson, C., & Sidman, M. (1993). Conditional discrimination and equivalence relations: control by negative stimuli. Journal of the Experimental Analysis of Behavior, 59, 333-347.

Kaminski, J., Call, J., & Fisher, J. (2004). Word learning in a domestic dog: evidence for fast mapping. Science, 304, 1682.

Kastak, C. R., & Schusterman, R. J. (2002). Sea lions and equivalence: expanding classes by exclusion. Journal of the Experimental Analysis of Behavior, 78, 449 465. doi:10.1901/jeab.2002.78-449.

Markman, E. M. (1987). How children constrain the possible meaning of words. New York: Cambridge University Press.

Markman, E. M. (1989). Categorization and naming in children. Cambridge: MIT Press.

Markman, E. M., & Hutchinson, J. E. (1984). Children's sensitivity to constraints on word meaning: taxonomic vs thematic relations. Cognitive Psychology, 16, 1-27.

Markman, E. M., & Wachtel, G. (1988). Children's use of mutual exclusivity to constrain the meanings of words. Cognitive Psychology, 20, 121-157.

McIlvane, W. J., Bass, R. W., O'Brien, J. M., Gerovac, B. J., & Stoddard, L. T. (1984a). Spoken and signed name and foods after receptive exclusion training in severe retardation. Applied Research in Mental Retardation, 5, 1-27.

McIlvane, W. J., & Dube, W. V. (2003). Stimulus control topography coherence theory: foundations and extensions. The Behavior Analyst, 26, 195-213.

McIlvane, W. J., Kledaras, J. B., Lowry, M. J., & Stoddard, L. T. (1992). Studies of exclusion in individuals with severe mental retardation. Research in Developmental Disabilities, 13, 509-532. doi: 10.1016 /0891-4222(92) 90047-A.

McIlvane, W. J., Kledaras, J. B., Munson, L. C., King, K. A., de Rose, J. C., & Stoddard, L. T. (1987). Controlling relations in conditional discrimination and matching by exclusion. Journal of the Experimental Analysis of Behavior, 48, 187-208.

McIlvane, W. J., Munson, L., & Stoddard, L. T. (1988). Some observations on control by spoken words in children's conditional discrimination and matching by exclusion. Journal of Experimental Child Psychology, 45, 472-495.

McIlvane, W. J., & Stoddard, L. T. (1981). Acquisition of matching-to-sample performances in severe mental retardation: learning by exclusion. Journal of Mental Deficiency Research, 25, 33-48.

McIlvane, W. J.. & Stoddard, L. T. (1985). Complex stimulus relations and exclusion in severe retardation. Analysis and Intervention in Developmental Disabilities, 5, 307-321.

McIlvane, W. J., Withstandley, J. K, & Stoddard, L. T. (1984b). Positive and negative stimulus relations in severely retarded individuals' conditional discrimination. Analysis and Intervention in Developmental Disabilities, 4, 235-251.

Merriman, W. E., Bowman, L. L., & MacWhinney, B. (1989). The mutual exclusivity bias in children's word learning. Monographs of the Society for Research in Child Development, 54, 1-129.

Mueller, M. M., Palkovic, C. M., & Maynard, C. S. (2007). Errorless learning: review and practical application for teaching children with pervasive developmental disorders. Psychology in the Schools, 44, 691-700.

Poulin-Dubois, D., & Graham, S. A. (2007). Cognitive processes in early word learning. In E. Hoff & M. Shatz (Eds.), Blackwell Handbook of language development (pp. 191-211). Oxford: Blackwell Publishing.

Rice, M. L., Buhr, J. C., & Nemeth. M. (1990). Fast mapping word-learning abilities of language-delayed preschoolers. Journal of Speech and Hearing Disorders, 55, 33-42. doi:10.1044 /jshd.5501.33.

Schmidt, A., Franco, M. G. 0., Loterio, L. S., & Gomes, G. F. (2016). Learning name-object relations after a single exclusion trial in 18- to 48-month-old children. The Psychological Record, 66, 53-63.

Serna, R. W., Foran, M. M., & Tran, S. L. (2016, May). Limitations to exclusion learning: The effects of difficult-to-discriminate stimuli on exclusion vs. outcome trials. Paper presented at the 42nd Annual Convention of the Association for Behavior Analysis International. Chicago, IL.

Sidman, M. (1985). Aprendizagem-sem-erros e sua importancia para o ensino do deficiente mental [Errorless learning and its relevance for teaching the mentally retarded], Psicologia, 11, 1-15.

Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.

Stoddard, L. T., de Rose, J. C., & McIlvane, W. J. (1986). Observafoes curiosas acerca do desempenho deficiente apos a ocorrencia de erros [Intriguing observations on poor performance after the occurrence of errors]. Psicologia, 12, 1-18.

Stoddard, L. T., & Sidman, M. (1967). The effects of errors on children 's performance on a circle-ellipse discrimination. Journal of the Experimental Analysis of Behavior, 10, 261-270.

Stromer, R. (1986). Control by exclusion in arbitrary matching-to-sample. Analysis and Intervention in Developmental Disabilities, 6, 59-72.

Stromer, R. (1989). Symmetry of control by exclusion in humans' arbitrary matching to sample. Psychological Reports, 64, 915-922.

Stromer, R., & Osborne, J. G. (1982). Control of adolescents' arbitrary matching-to-sample by positive and negative stimulus relations. Journal of the Experimental Analysis of Behavior, 37, 329-348.

Swingley, D. (2010). Fast mapping and slow mapping in children's word learning. Language Learning and Development, 6, 179-183.

Tardif, T. (1996). Nouns are not always learned before verbs: evidence from Mandarin speakers' early vocabulary. Developmental Psychology, 32, 492-504.

Tomonaga, M. (1993). Tests for control by exclusion and negative stimulus relations of arbitrary matching to sample in a "symmetry-emergent" chimpanzee. Journal of the Experimental Analysis of Behavior, 59, 215-229.

Vincent-Smith, L., Bricker, D., & Bricker, W. (1974). Acquisition of receptive vocabulary in the toddler-age child. Child Development. 45, 198-193.

Waxman, S., Fu, X., Arunachalam, S., Leddon, E., Geraghty, K., & Song, H. (2013). Are nouns learned before verbs? infants provide insight into a long standing debate. Child Development Perspectives, 7, 155-159.

Wilkinson, K. M., de Souza, D. G., & McIlvane, W. J. (2000). Origens da exclusao [Origins of exclusion]. Temas em Psicologia, 8, 195-203.

Wilkinson, K. M., Dube, W. V., & McIlvane, W. J. (1998). Fast mapping and exclusion (emergent matching) in developmental language, behavior analysis and animal cognition research. The Psychological Record, 48, 407-422.

Wilkinson, K. M., & McIlvane, W. J. (1994). Stimulus organization and learning by exclusion: a preliminary experimental analysis. Experimental Analysis of Human Behavior Bulletin, 12, 21-25.

Wilkinson, K. M., & McIlvane, W. J. (1997). Blank comparison analysis of emergent symbolic mapping by young children. Journal of Experimental Child Psychology, 67, 115-130.

Wilkinson, K. M., Rosenquist, C., & McIlvane, W. J. (2009). Exclusion learning and emergent symbolic category formation in individuals with severe language impairments and intellectual disabilities. The Psychological Record, 59, 187-206.

Thais Arantes Ribeiro (1) * Tamiris Poletini Gallano (1) * Debora de Hollanda Souza (1) * Deisy das Gramas de Souza (1)

(1) Institute Nacional de Ciencia e Tecnologia sobre Comportamento, Cognigao e Ensino (INCT-ECCE), Universidade Federal de Sao Carlos, Rodovia Washington Luis, Km 235, Sao Carlos, SP, Brazil 13.561-381

Caption: Fig. 1 The experimental sequence across conditions and phases. For each condition, the baseline phase established three auditory-visual relations with meaningful stimuli. Phases 3 and 4 introduced exclusion (nonreinforced for the first trial and reinforced for subsequent trials) and learning outcome probes, each with one pair of undefined stimuli, interspersed among baseline trials. Noun and Adjective relations were counterbalanced as the first or the second condition for participants exposed to both conditions. Some participants were exposed only to the first condition (either Noun or Adjective)

Caption: Fig. 2 Number of exposures to each training block (1 to 5) of baseline (Phase 2) in the Noun and Adjective conditions (white and black bars, respectively) for each participant. The two rightmost bars show average data and the standard error of the mean for the two conditions. Participants' data are ordered according to the conditions to which they were exposed: Noun condition first, only Noun, Adjective condition first, and Adjective only. * p < 0.05. ** p< 0.005

Caption: Fig. 3 Average proportion of correct responses per participant on exclusion trials (non-reinforced and reinforced), and Control Probes 1 and 2 (Phases 3 and 4). Upper panel: Noun condition. Bottom panel. Adjective condition

Caption: Fig. 4 Average proportion of correct responses for all participants on learning outcome probes (Phases 3 and 4). For each probe, the white bar represents results after the first exclusion probe, and the black bar represents the results after all exclusion trials (non-reinforced and reinforced). The two rightmost bars (All) show the average proportion of correct responses considering the three learning probes together after the first exclusion probe (white bar) and after all exclusion trials (black bar). Data from the Noun condition is displayed in the upper panel, and from the Adjective condition in the bottom panel. * p < 0.05. ** p < 0.005. *** p< 0.0001
Table 1 Participants' identification, gender, chronological age,
language-age equivalent (PPVT-R), and order of exposure to
experimental conditions

Participant   Gender   Chronological Age   Language-Age
                         (year-month)       Equivalent
                                             (PPVT-R)

P01             F           2-0/2-4            2-3
P02             M           2-2/2-6            2-4
P03             M           2-1/2-5            2-2
P04             M           2-1/2-3            2-1
P05             M           2-1/2-5            2-1
P06             M           2-0/2-4            2-1
P07             M           2-1/2-2            2-2
P08             M           2-3/2-4            2-6
P09             F           2-4/2-6            2-1
P10             F           2-3/2-5            2-6
Pll             F           2-2/2-5            2-1
P12             M           2-2/2-4            2-6
P13             M           2-2/2-4            2-6
P14             M           2-3/2-7            2-0
P15             F           2-0/2-4            2-3
P16             F           2-1/2-5            2-1
P17             F           2-5/2-6            2-3
P18             M           2-3/2-5            2-2
P19             F           2-4/2-6            2-2
P20             F           2-5/2-8            2-2
P21             F           2-2/2-6            2-0

Participant   Experiments Conditions

                 1st         2nd

P01             Noun      Adjective
P02             Noun      Adjective
P03             Noun      Adjective
P04             Noun      Adjective
P05             Noun      Adjective
P06             Noun      Adjective
P07             Noun         --
P08             Noun         --
P09           Adjective     Noun
P10           Adjective     Noun
Pll           Adjective     Noun
P12           Adjective     Noun
P13           Adjective     Noun
P14           Adjective     Noun
P15           Adjective     Noun
P16           Adjective     Noun
P17           Adjective      --
P18           Adjective      --
P19           Adjective      --
P20           Adjective      --
P21           Adjective      --

Note: Gray cells indicate participants exposed to a single condition.
Chronological age is shown for the age at the beginning
and end of the experiment.

Language-age equivalent is from the beginning of the experiment

Table 2 Defined (D) and undefined (U) sample stimuli and
comparisons designated as correct in conditions 1 and 2

Condition 1: Name-picture matching

Sample Stimuli        Comparison
                  Designated Correct

Defined Stimuli (D)

/bola/ DN1               DN1

/moto/ DN2               DN2

/aviao/ DN3              DN3

Undefined Stimuli (U)

/beva/ UN1               UN1

/fapi/ UN2               UN2

/nopa/ UN3               UN3

/mupa/ UN4               UN4

(No word)                UN5

(No word)                UN6

Condition 2: Adjective-picture matching

Sample Stimuli        Comparison
                  Designated Correct

/brava/ DA1              DA1

/feliz/ DA2              DA2

/triste/ DA3             DA3

/fobs/ UA1               UA1

/piva/ UA2               UA2

/mipa/ UA3               UA3

/nafu/ UA4               UA4

(No Adjective)           UA5

(No Adjective)           UA6

Note: The numbers identify individual stimuli

Table 3 Stimuli used on exclusion, Control, and
Learning Outcome probes for the Noun condition

Probes                        Sample     Comparisons

                                          S-     S-     S+

Phase 3 (Relation 1)

Exclusion                   /beva/UN1    DN1     M     UN1
Learning Outcome Probe A    /beva/UN1    UN3     M     UN1
Learning Outcome Probe B    /beva/UN1    DN3    UN5     M
Learning Outcome Probe C    /nopa/UN3    DN2    UN1     M
Control 1                   /bola/DN1    UN1     M     DN1
Control 2                   /beva/UN1    DN1    DN2     M

Phase 4 (Relation 2)

Exclusion                   /fapi/UN2    DN2     M     UN2
Learning Outcome Probe A    /fapi/UN2    UN4     M     UN2
Learning Outcome Probe B    /fapi/UN2    DN1    UN6     M
Learning Outcome Probe C    /mupa/UN4    DN3    UN2     M
Control 1                   /aviao/DN3   UN2     M     DN3
Control 2                   /fapi/UN2    DN2    DN3     M

Note: The S+ column indicates the selection response that would
be consistent with exclusion responding, or with learning of the
name-picture relation, for each trial. D defined stimuli,
U undefined stimuli, A adjective, M mask

Table 4 Stimuli used on Exclusion, Control, and
Learning Outcome for the Adjective condition

Probes                         Sample      Comparisons

                                            S-     S-     S+

Phase 3 (Relation 1)

Exclusion                    /foba/ UA1    DA1     M     UA1
Learning Outcome Probe A     /foba/ UA1    UA3     M     UA1
Learning Outcome Probe B     /foba/ UA1    DA3    UA5     M
Learning Outcome Probe C     /mipa/ UA3    DA2    UA1     M
Control 1                   /brava/ DA1    UA1     M     DAI
Control 2                    /foba/ UA1    DA1    DA2     M

Phase 4 (Relation 2)

Exclusion                    /piva/ UA2    DA2     M     UA2
Learning Outcome Probe A     /piva/ UA2    UA4     M     UA2
Learning Outcome Probe B     /piva/ UA2    DA1    UA6     M
Learning Outcome Probe C     /nafu/ UA4    DA3    UA2     M
Control 1                   /triste/ DA3   UA2     M     DA3
Control 2                    /piva/ UA2    DA2    DA3     M

Note: The S+ column indicates the selection response that would
be consistent with exclusion responding, or with learning of the
name-picture relation, for each trial. D defined stimuli,
U undefined stimuli, A adjective, M mask

Table 5 Number of selections of undefined stimuli per opportunity
on the first exclusion probe for each relation in the Noun and
Adjective conditions

Condition   Undefined   Noun    Adjective
            stimuli     first   first

Noun        Beva-UNl    6/6     8/8
            Fapi-UN2    5/6     8/8
            Subtotal
Adjective   Foba-UAl    3/6     8/8
            Piva-UA2    6/6     6/8
            Subtotal
            Total

Condition   Undefined   Noun    Adjective   Total
            stimuli     only    only

Noun        Beva-UNl    1/2     --          15/16
            Fapi-UN2    2/2     --          15/16
            Subtotal                        30/32
Adjective   Foba-UAl    --      5/5         16/19
            Piva-UA2    --      3/5         15/19
            Subtotal                        31/38
            Total                           61/70

Note: There was only one non-reinforced exclusion probe
for each relation per participant; therefore, the number
of opportunities for the first trial was equal to the number
of participants. The first two columns refer to participants
exposed to the two conditions; 14 participants completed both
conditions and seven completed only one condition (two
completed the Noun and five completed the Adjective condition).
The denominator indicates the total number of opportunities

Table 6 Correct responses for each word-picture relation on
Learning Outcome probes as a function of the number of
exclusion trials (1, 4, 7)

Number of exclusion    Number of      Learning probes
trials (a)             participants
                                      Probe    Probe   Probe
                                      A        B       C
                                      N(%)     N(%)    N (%)

Noun
  Beva
    1                  16             11(69)   3(19)   4(25)
    4                  15             10(67)   3(20)   8(53)
    7                  13             10(77)   2(15)   3(23)
    10                 13             11(85)   2(15)   2(15)
  Fapi
    1                  16             7(44)    1 (6)   3(19)
    4                  16             15(94)   1 (6)   6(34)
    7                  16             10(63)   4(25)   7(44)
    10                 15             10(67)   5(33)   9(60)
Adjective
  Foba
    1                  19             8(42)    5(26)   6(32)
    4                  19             12(63)   5(26)   5(26)
    7                  17             12(71)   4(24)   0(0)
    10                 17             13(76)   5(29)   6(35)
  Piva
    1                  19             16(84)   7(37)   5(26)
    4                  16             12(75)   8(50)   4(25)
    7                  13             12(92)   3(23)   3(23)
    10                 13             9(69)    4(31)   3(23)

Note: The number of participants varied because participants
who reached the learning criterion for the three learning
outcome probes left the experiment without exposure
to additional probes.

(a) Only the first trial was truly an "exclusion"
(non-reinforced) probe; the subsequent trials presented
the same arrangement of stimuli, but selection responses
produced differential consequences

Table 7 Number of exposures to exclusion trials (a)
(1, 4, 7, or 10) after which a relation was established

                Noun Condition

            Beva             Fapi

        All    Probe A   All    Probe A

Noun First

P1       No       1       No      10
P2       No       7       No       7
P3       No       7       No      No
P4       No      No       No      No
P5       No       7       No      No
P6       4        1       No      No
P7       4        4       10       4
P8       No       1       No      10

Adjective First

P9       No       1       No       4
P10      1        1       7        1
P11      10       4       7        7
P12      No       1       No      No
P13      No       1       No       1
P14      No       1       No       1
P15      No      10       4        4
P16      No       4       No       4
P17      --      --       --      --
P18      --      --       --      --
P19      --      --       --      --
P20      --      --       --      --
P21      --      --       --      --
Total   4/16    15/16    4/16    11/16

             Adjective Condition

            Foba             Piva

        All    Probe A   All    Probe A

Noun First

P1       10       4       1        1
P2       No      No       No       1
P3       No      No       1        1
P4       No       4       1
P5       10      10       1
P6       No       7       No      No
P7       --      --       --      --
P8       --      --       --      --

Adjective First

P9       No       1       10       1
P10      10      10       4        4
P11      No       4       No       1
P12      4        1       4        1
P13      No       1       10       1
P14      No      10       No      No
P15      No       1       No       7
P16      No       1       No       1
P17      4        1       4        1
P18      No      No       No       1
P19      No       4       No      No
P20      No      No       No      No
P21      No       7       No       4
Total   5/19    15/19    9/19    15/19

Note: The numbers indicate after how many
exclusion trials the criterion was met.

"No" indicates that the participant did not achieve
the learning criterion after 10 trials. Dashes indicate
that the participant was not exposed to the specific probe.
Gray cells indicate participants who achieved the learning
criterion when the three learning probes for each relation
are considered.

(a) Only the first trial was truly an "exclusion"
(non-reinforced) probe; the subsequent trials presented
the same arrangement of stimuli, but selection responses
produced differential consequences

Table 8 Responding on learning probes across studies: participant ages,
number of participants, and percentage of responses consistent with
learning on Probes A, B, and C

References            Ages            N

Present               24-29 months    19
study--Noun

Present               24-29 months    16
study--Adjective

Schmidt et al.        18-30 months    40
(2016)

Costa (2009)          24-36 months    60

Domeniconi et al.     25-32 months    6
(2007)

Wilkinson &           39-60 months    8
McIlvane (1997)

Costa, Wilkinson,     41-71 months    17
McIlvane & de Souza
(Exp. 1, 2001)

Costa et al. (2013)   59-70 months    8

Costa et al.          73-140 months   35
(Exp. 2, 2001)

References            Percentage

                      Probe type

                      Probe A (%)   Probe B (%)   Probe C (%)

Present               68            15            35
study--Noun

Present               72            34            28
study--Adjective

Schmidt et al.        --            13            45
(2016)

Costa (2009)          61            48            --

Domeniconi et al.     --            33            67
(2007)

Wilkinson &           --            25            50
McIlvane (1997)

Costa, Wilkinson,     --            6             12
McIlvane & de Souza
(Exp. 1, 2001)

Costa et al. (2013)   81            75            --

Costa et al.           --            23            17
(Exp. 2, 2001)

Note: Adapted from Costa et al. (2014). The studies
were selected based on availability of data and types of
learning probes used. The symbol "--" indicates that the
probe type was not used in that particular study
COPYRIGHT 2017 Springer
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:ORIGINAL ARTICLE
Author:Ribeiro, Thais Arantes; Gallano, Tamiris Poletini; Souza, Debora de Hollanda; de Souza, Deisy das Gr
Publication:The Psychological Record
Article Type:Report
Date:Sep 1, 2017
Words:14186
Previous Article:Erratum to: the effect of conditional discrimination reversals with SMTS and DMTS on reorganization of equivalence classes.
Next Article:Omission training results in more resurgence than alternative reinforcement.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters