Effect of Social Interactions on Song Learning in Zebra Finches

Effect of Social Interactions on Song Learning in Zebra Finches A thesis submitted in partial fulfillment of the requirements for graduation with Honors in Psychology 2021 ii Table of Contents Acknowledgements ............................................................................................................ iv Abstract ............................................................................................................................... v List of Figures .................................................................................................................... vi Introduction ......................................................................................................................... 1 Role of Peer Interactions on Learning During Development ......................................... 2 Parallels between Humans and Zebra Finches................................................................ 4 Song Learning in Zebra Finches ..................................................................................... 5 Role of Peers and Environment in Song Performance and Learning in Zebra Finches . 6 Social Facilitation and Inhibition of Song Development ................................................ 8 Present Study ................................................................................................................ 10 Materials and Methods ...................................................................................................... 10 Subjects ......................................................................................................................... 10 Procedures and Measurements ...................................................................................... 11 Experimental Timeline.................................................................................................. 11 Song Recording and Analysis ....................................................................................... 12 Statistical Analyses ....................................................................................................... 13 Results ............................................................................................................................... 14 Exploratory Analyses .................................................................................................... 15 Discussion ......................................................................................................................... 22 Social Inhibition vs. Social Facilitation ........................................................................ 23 Ecological Explanations................................................................................................ 24 Ecological and Instrumentation Limitations ................................................................. 26 Implications and Future Directions ............................................................................... 27 Conclusion ........................................................................................................................ 28 Bibliography ................................................................................................................. 29 iii Acknowledgements iv Abstract Humans are a social species whose learning benefits from peer interaction. Benefits include heightened language ability, better problem-solving capabilities, and improved academic performance. However, the effects of social isolation on learning are ambiguous for humans and other non-human species. Zebra finches (Taeniopygia guttata) share an array of similar neural and social mechanisms as humans concerning song and language learning. Through a longitudinal study, we analyzed songs from 19 male zebra finches from ~60 days post-hatch (dph) until ~100 dph. We assigned males to either a group-housing or social isolation condition. We predicted that song accuracy for both groups would increase, but that song accuracy would be higher for the group-housed condition. Song accuracy for both groups did not improve over time and the grouphoused condition did not perform significantly better than the isolation group. Due to the poor song accuracy, we ran exploratory analyses by comparing siblings’ songs to each other. The exploratory analysis revealed that social interactions inhibit song development. Ecological discrepancies, individual differences, and fraternal inhibition may be contributing factors to the poor observed song performance. Future research must continue the utilization of animal models to determine if social isolation is detrimental to learning. v List of Figures Figure 1: Spectrograms of a tutor with two of its pupils, both siblings, in the grouphousing and social isolation conditions. ........................................................................ 17 Figure 2: Median SI Scores Across Groups ................................................................... 18 Figure 3: Median SI Scores Across Time Bins and Between Groups ........................... 19 Figure 4: Median SI Scores Relative to Siblings Between Groups ............................... 20 Figure 5: Comparison of Group-housed Condition Median SI Scores Relative to Tutor and Siblings .................................................................................................................... 21 vi Introduction Since April 2020, the COVID-19 pandemic has resulted in the social isolation of 1.8 billion students worldwide from their peers (United Nations Sustainable Development [UNSD], 2020). This is problematic because children who interact with peers have better problem-solving abilities, better language skills, and work more effectively in a group than those without peer interaction (Bolhuis et al., 2010; Chib et al., 2018; Coolahan et al., 2000; Fawcett et al., 2005). Fromkin et al. (1974) have linked social isolation to abnormal cognitive development, but there are few studies on the effect of human isolation on learning ability. Cases of extreme neglect, however, show that social isolation can cause serious developmental problems. For instance, Fromkin et al. (1974) provided the example of Genie, who was deprived of human interaction during the majority of her childhood and showed social and language difficulties, such as the inability to comprehend grammatical structures. Persistent intelligence deficits have been observed in orphaned Romanian children deprived of social interactions (Chugani et al., 2001), further highlighting the crucial role of early-life social experience on cognitive development (Chugani et al., 2001) and learning (Fromkin et al., 1974). Role of Peer Interactions on Learning During Development The Early Head Start (EHS) program demonstrates the importance of early-life experiences for learning and development in young children (Love et al., 2005). By providing early learning opportunities, the EHS program prepares children of low socioeconomic status for kindergarten. Children who participated in the program displayed greater cognitive and language development than children who did not participate (Love et al., 2005). However, these cognitive gains are not limited to specialized programs but may instead be the result of interactions with peers (Coolahan et al., 2000). Peer interactions in preschool children can benefit learning. Preschool children who engaged with their peers showed greater long-term academic achievement and motivation than those who did not (Coolahan et al., 2000). Furthermore, children who interacted with peers frequently displayed more attentiveness than children who engaged with their peers infrequently (Coolahan et al., 2000). This suggests that peer engagement has a positive relationship with learning and school performance. Specifically, positive play and peer engagement corresponded to a 92% increase in classroom attentiveness (Coolahan et al., 2000). However, school closures in 2020 due to COVID-19 have halted peer interactions, therefore inhibiting the benefits of peer interactions in learning environments (Chib et al., 2018; Coolahan et al., 2000; Fawcett et al., 2005). Interactions with peers not only improves attention and motivation in the classroom but also serves as a crucial link in learning (Fawcett & Garton, 2005). The presence of peers can help bridge differences in problem-solving abilities among students (Fawcett & Garton, 2005), promote cognitive conflict resolution, and reinforce positive behaviors 2 (Wafters & Ginns, 1997). Yet, Hanushek et al. (2003) found no evidence to suggest that peer interactions improve academic achievement. However, as the learning environment is not dissociable from the effect of peers, the relationship between the two factors and their effect on learning outcomes remains unclear (Hanushek et al., 2003). Also, the benefits of learning may average out over time, thus making it unclear if peer interactions have a lasting effect on learning (Hanushek et al., 2003). On the other hand, Fawcett and Garton’s (2005) findings suggest that interactions with peers improves learning retention. For example, in problem-solving tasks, six and seven-year-old students who worked in pairs and freely exchanged ideas outperformed those who did not collaborate. Furthermore, children with low problem-solving ability paired with high-ability children demonstrated lasting gains in problem-solving skills, although improvements were limited to those who were required to explain their strategies to their partners. These results further highlight the importance of social interactions on learning (Fawcett & Garton, 2005). Peer interactions not only aids in learning outcomes for young children (Fawcett & Garton, 2005) but also assists in the learning of young adults (Bakar & Orji, 2018). Bakar and Orji (2018) found that college students who alternated between roles as tutors and learners reported more motivation to learn and outperformed students who were taught only by instructors. Although interactions with peers improves learning outcomes (Bakar & Orji, 2018; Fawcett & Garton, 2005), the effect of social isolation on learning outcomes is unclear, so we utilized an animal model to investigate this uncertainty. 3 Parallels between Humans and Zebra Finches Zebra finches are a highly social songbird species (Brainard & Doupe, 2002; Immelmann, 1969; Price, 1979) whose song development depends on social interactions (Caruso-Peck & Goldstein, 2019) and exposure to a tutor song (Immelmann, 1969). Both humans and zebra finches are vocal learners, meaning they can modify their vocal output based on individualized experience (Brainard & Doupe, 2013; Immelmann, 1969; Martins & Boeckx, 2020), a trait shared by only eight other groups of animals (Martins & Boeckx, 2020). Zebra finches demonstrate sophisticated motor learning through auditory stimulation and feedback while also relying on social reinforcement and social interactions to acquire correct vocalizations (Brainard & Doupe, 2002). Furthermore, zebra finches not exposed to template songs develop abnormal songs similar to the effect of experiential deprivation in humans during language development (Immelmann, 1969; Price, 1979). In addition to requiring a live tutor (Immelmann, 1969), zebra finches rely on specialized brain regions to guide their song learning (Brainard & Doupe, 2013). Humans and zebra finches rely on auditory feedback via similar cortico-striatal-thalamocortical loops that subserve the learning of vocalizations (Brainard & Doupe, 2013). Furthermore, deafening juvenile zebra finches halts song development and leads to song deterioration in males that have either begun practicing or have acquired song, demonstrating that learning depends on coordinating vocalizations with auditory feedback (Brainard & Doupe, 2013). Similarly, human children with congenital or acquired deafness produce aberrant vocalizations (Brainard & Doupe, 2013). Moreover, cooling of the HVc song nuclei in zebra finches, which has analogous functions in song as Broca’s area in human 4 speech, elicits similar vocal “stretching” as cooling Broca’s area in humans (Long et al., 2016; Long & Fee, 2008). Both humans and zebra finches display sensitive periods during which sensory experience is necessary for normal language and song development (Doupe & Kuhl, 1999). As a result of these similar neurological mechanisms, which correlate with song behavior (Brainard & Doupe, 2002), researchers have utilized zebra finches as a robust model organism for the study of perceptual and sensory learning processes (Brainard & Doupe, 2013). Zebra finch song performance is subserved by discrete brain circuits that are also present in mammals (Brainard & Doupe, 2013) and provides researchers with a quantifiable behavior tractable to experimental manipulations (Brainard & Doupe, 2002). Furthermore, zebra finches provide researchers with experimental control by virtue of their small size (Brainard & Doupe, 2002). Also, their rapid biological development allows researchers to determine experimental outcomes with minimal time constraints (Brainard & Doupe, 2002). Song Learning in Zebra Finches Zebra finches are a “closed-ended” estrildid species that display overlapping sensitive periods during song acquisition (Brainard & Doupe, 2002; Immelmann, 1969). Males typically begin encoding their tutor’s song after fledging at approximately 20 days post-hatch (dph) which can continue until males reach sexual maturity at approximately 90 dph (Immelmann, 1969). During this sensory learning period, pupils encode a “template” song learned from an adult male, which depends on auditory stimulation (Brainard & Doupe, 2002; Immelmann, 1969; Price, 1979). Males practice their song 5 during the sensorimotor period which typically extends from 35 to 90 dph (Immelmann, 1969; Price, 1979). Extensive vocal practice allows males to develop a mature song that corresponds to the encoded “template” (Brainard & Doupe, 2013). In the absence of a tutor, zebra finches will integrate the calls of females or the songs of other species into their repertoire (Price, 1979). Immelmann’s (1969) study on “isolates” mapped the developmental course of song acquisition and demonstrated that for normal song development, juveniles must both hear and interact with a tutor. As males fledge and transition to the sensorimotor phase, their song remains plastic, or marked by variability, until “crystallization” occurs at approximately 90 dph, which coincides with sexual maturity (Price, 1979). Crystallization is marked by song stereotypy (Immelmann, 1969) and songs composed of complex and repeated motifs of discrete syllables (Brainard & Doupe, 2013). However, isolates’ songs can remain plastic past sexual maturity (Price, 1979) and tend to lack the complexity demonstrated in the song of tutored males (Price, 1979). Role of Peers and Environment in Song Performance and Learning in Zebra Finches Both the singing behavior (Jesse & Reibel, 2012) and the song learning process (Carouso-Peck & Goldstein, 2019) are sensitive to the social environment (Bölting & von Engelhardt, 2017) and social cues (Carouso-Peck & Goldstein, 2019) as shown by the frequency and duration of the focal male’s songs in the company of conspecifics and nonconspecifics (Jesse & Reibel, 2012). Carouso-Peck and Goldstein (2019) found that 6 males who received feedback from computer-generated female images produced songs analogous to the songs of their tutor, while males that were not provided feedback produced songs unique from those of their tutors. These studies suggest that the presence of conspecifics and non-conspecifics has a marked effect on the development of songs in zebra finches during adolescence (Carouso-Peck & Goldstein, 2019; Jesse & Reibel, 2012). Despite strong evidence showing that tutoring is critical for the normal development of songs (Immelmann, 1969; Jones et al., 1996; Price; 1978), less is known about the influence of same-age conspecifics on learning (Jones et al., 1996). Genetic brothers transmit songs to siblings isolated from tutors demonstrating that learning occurs through horizontal transmission (Derégnaucourt & Gahr, 2013). For example, male zebra finches housed in small groups and isolated from tutors at different periods before sexual maturity shared syllables and made more similar songs than males reared with tutors (Jones et al., 1996). This suggests the presence of same-age conspecifics affects song composition. Additionally, Jesse and Riebel (2012) found that the presence of both males and females facilitated song production in focal males; however, their study ran for only 24 hours and did not find any difference in the accuracy of the song in males who interacted with same-age conspecifics and non-conspecifics. Together, these findings suggest song learning is driven by factors beyond tutoring (Derégnaucourt & Gahr, 2013). Derégnaucourt and Gahr’s (2013) study implies that siblings serve as protection against the development of aberrant songs when learning is disrupted by experiential deprivation in the form of the absence of a tutor. Yet, it is still uncertain how the presence of same-age conspecifics affects learning in males that have 7 encoded a template song during the sensory learning period (cf. Immelmann, 1969; Price, 1979). An additional complication is that Jesse and Riebel (2012) did not find any differences in the accuracy of songs in males that interacted with same-age conspecifics or non-conspecifics. The lack of convergence between these findings suggests that the role peers play in the development of mature songs is ambiguous and will require further research to elucidate (Derégnaucourt & Gahr, 2013; Jones et al., 1996; Jesse & Riebel, 2012). Social Facilitation and Inhibition of Song Development The literature on the social determinants of song development is further complicated when considering the influences of oblique transmission of syllables (Jones et al., 1996) and the convergence of improvised syllables on adult songs (Price, 1979). Males can horizontally transmit the syllables of non-related tutors to conspecifics through social interactions even after males form a template of their related tutor’s song (Jones et al., 1996). Rather than social interactions facilitating song development (Carouso-Peck & Goldstein, 2019), social interactions can also inhibit the development of accurate copies of a juvenile’s tutor’s song (Jones et al., 1996). Convergence of improvised syllables in group-housed males during the sensorimotor phase may contribute to poor tutor song imitation as well (Price, 1979). Juvenile males in group housing generate spontaneous syllables that converge across the males and become fixed in their repertoires (Jones et al., 1996; Immelmann, 1969; Price, 1979). Across multiple studies, males housed with conspecifics during the sensorimotor stage shared more syllables in their adult songs with conspecifics than with 8 their tutors (Jones et al., 1996; Price, 1979). Jones et al. (1979) found group-housed juvenile males shared more than half of their syllables with conspecifics while sharing less than half of their syllables with their tutor. Social inhibition of song development might be accounted for by the convergence of improvised syllables during the sensorimotor period (Jones et al., 1996; Price, 1979). Social inhibition might also explain the null findings of Jesse & Riebel (2012) as social interactions increased the frequency of singing behavior of group-housed conspecifics, yet song accuracy did not improve. Though social interactions between conspecifics increased the frequency of singing behavior, increased singing may have led to increased inhibition of accurate song development as well (Jones et al., 1996). The extant literature is rife with conflicting findings on the role of social interaction on song development. Another complicating factor for interrogating the effect of social interaction on song development is fraternal inhibition (Tchernichovski & Nottebohm, 1998), which posits that interactions between juvenile siblings during the sensorimotor period inhibited the development of accurate copies of the tutor’s song. Tchernichovski and Nottebohm’s (1998) findings conflict with Derégnaucourt and Gahr’s (2013) conclusion that siblings serve as protective factors against abnormal song development. Social facilitation has been observed in juveniles when males received contingent feedback during singing behavior (Carouso-Peck & Goldstein, 2019), though social inhibition of song development has been observed as well (Jones et al., 1996; Price, 1979; Tchernichovski & Nottenbohm, 1998). Together, these findings are consistently contradictory, indicating that the role of social interactions on song development remains largely unknown. 9 Present Study Past research on the effects of social isolation on learning is unclear for both humans and zebra finches. The purpose of this study is to determine how social interactions with male conspecifics during the sensorimotor phase affects song learning. We investigated this ambiguity by assigning males into either a group-housing or social isolation condition. Research has shown that zebra finch songs will mature over time after they have developed a template song (Brainard & Doupe, 2013). (1) Therefore, we expect song accuracy to increase over time for both groups. Additionally, research has shown that the song learning process is highly sensitive to the social environment and social cues (Bölting & von Engelhardt, 2017; Carouso-Peck & Goldstein, 2019; Jesse & Reibel, 2012). So, if social interactions facilitate learning, (2) the learning trajectory for males housed with conspecifics should be higher than that of the isolated males such that the group-housing condition will produce more accurate copies of their tutors’ songs over fewer days than males isolated from conspecifics. Materials and Methods Subjects All procedures were approved by the Whitman College Institutional Animal Care and Use Committee for experiments involving vertebrate animals. Nineteen male zebra finch (Taeniopygia guttata) subjects were selected from five different breeding cages located at Whitman College. Two subjects were excluded from analysis due to incomplete data. Five subjects in the isolation condition and 12 in the group-housing condition remained for final analysis. Juvenile birds were separated from their tutor at 40 10 dph (M: 42.35 dph; range: 39 - 59 dph). A single bird (Orange175) was the sole pupil of a tutor; all other tutors raised 2-6 brothers in the current study. Subjects were housed in a climate-controlled room on a 13:11 light/dark cycle including 30 minutes of dawn and dusk lighting, except for isolation chambers that did not have this feature. Isolated males were weighed every two weeks to ensure the health of the subjects. Birds were given unlimited access to food, grit, and water; nutritional supplements (e.g., Calci-boost, vegetables, cuttlebone, chopped hard-boiled eggs, and spray millet); and environmental enrichment (e.g., perches of different sizes, swings, and water baths). Procedures and Measurements Experimental Timeline All birds were reared in a nest with a single male cock (tutor) and female hen in a larger room housing ~100 other finches. Opaque dividers were placed between breeding cages to visually isolate cages from one another and to ensure that pupils had social interactions solely with their tutor, hen, and clutchmates. All reasonable attempts were made to match male siblings across conditions. To determine the effect of conspecifics on song accuracy, we separated fledged males from their tutors and randomly assigned them to one of two housing conditions: social isolation or group-housing. Males were removed from their home nest after 40 dph, which allowed subjects sufficient time to memorize a template song from their tutor during the sensory phase (Immelmann, 1969). Following assignment to housing conditions, males were recorded weekly until approximately 100 dph. Isolated finches remained in visual and aural isolation for the entirety of the study 11 until reintroduction to the colony. Birds in the group-housing condition were only removed on recording days and were housed singly for ~24 hours. At ~100 dph, subjects were reintroduced to the colony. The tutor of each bird was recorded in a sound attenuation chamber and was returned to his home cage after song files were acquired. Song Recording and Analysis Recordings occurred every seven days and song accuracy were compared across groups at 60 (range: 57 - 62), 70 (range: 67-72), 80 (range: 77 - 82), and 100 (range: 97 102) dph. Songs were collected across the entirety of the experiment by housing animals singly in sound attenuation chambers. All reasonable attempts were made to record a given bird using the same microphone and recording devices/settings, with some differences in the quality of recordings between time points. Sounds were recorded using Shure SM58 microphones, digitized using a U-Phoria UMC4040HD (44.1 kHz sampling rate, 24-bit depth) and recorded using Sound Analysis Pro (SAP) 2011 software (Day et al., 2019; Tchernichovski et al., 2000). Only songs sung between 10 am and 12 pm were analyzed. All song recordings were undirected (i.e., not directed towards a female). 25 pupil bouts were compared to five tutor motifs in a given time bin across the experimental timeline. Song bouts were randomly selected if there were more than 25 in the time window. While selecting for song bouts and motifs, we were blinded to the subjects’ names and conditions to reduce researcher bias. The Similarity Index (Mandelblat-Cerf & Fee, 2014), a Matlab-based song similarity analysis program, quantified how well pupils imitated their tutor (‘Imitation’ sub-score). The Imitation score has a maximum value of 1 (a perfect match to the tutor) and a minimum value of 0 12 (no match to the tutor). Imitation is a product of Acoustic Similarity (the extent to which the vocal sound signal matches between tutor and pupil) and Sequence Similarity (the extent to which the order of syllable production matches). Analyses were conducted using the recommendations of the authors in the user manual. The Similarity Index compared segments of bouts with the spectral features of motifs to derive a similarity score (SI) which was used as the basis for statistical tests. Canonical and non-canonical renditions of motifs were included in the analyses to accurately reflect the variability in the song in both tutor motifs and pupil bouts. An effort was made to remove all non-integrated calls and noise when extracting motifs and bouts for analysis. Statistical Analyses Prior to conducting any analyses, we extracted the required pupil bouts and tutor motifs. Once extracted, we inputted the songs into the similarity index software (Mandelblat-Cerf & Fee, 2014), which assessed how well pupils copied their tutor’s song. After all songs were analyzed, we used the resulting similarity index (SI) scores for all statistical analyses. Due to incomplete data and a small sample size, we utilized nonparametric tests for all statistical analyses. First, we assessed changes in song accuracy across time by binning the average SI scores of early time points into a single variable (Days 60 and 70; 57-72 dph) and later time points into a single variable (Days 90 and 100; 87-102 dph). This manipulation allowed us to use the Wilcoxon Signed Rank test to determine if learning had occurred across time. To determine if the housing condition had an effect on learning, we created a “total learning” variable which was the difference of 13 days 60 and 70 from days 90 and 100 and then applied a Mann-Whitney U test. Also, we explored the extent to which males learned more accurate copies of their tutors or siblings’ songs by computing the SI scores of siblings and comparing these scores to the tutor SI scores using a Mann-Whitney U test. We next compared SI scores relative to siblings between groups to determine which group learned more accurate copies of their siblings’ songs using a Mann-Whitney U test. Finding a meaningful difference between groups, we then focused on the group with the greater scores and compared SI scores relative to siblings to that of SI scores relative to tutors using a Mann-Whitney U test. Results We first established inter-rater reliability. Raters were trained on proper song analysis procedures and were provided blinded sets of song files. Only three birds had complete data sets among all time bins. These three birds were used to obtain inter-rater reliability (𝛂 = .950) which demonstrated sufficient inter-rater reliability for further analysis. To ensure the SI scores adequately represented song accuracy, we qualitatively assessed spectrograms across ages and determined that the SI scores were invalid representations of song accuracy (Figure 1). This suggests that our measure was not appropriately assessing song similarity, which makes it unclear if the resulting SI scores are indeed accurate representations of song accuracy. To determine if song accuracy improved with age, we averaged SI scores from days 60 and 70 (57-72 dph) and days 90 and 100 (87-102 dph). The median SI scores at days 60 and 70 (Mdn = 0.322, IQR = ± 0.125) did not significantly differ from the 14 median SI scores at days 90 and 100 (Mdn = 0.282, IQR = ± 0.146, z = -1.511, p = .131, r = 0.456), which does not support our hypothesis (1) that song accuracy would increase with age (Figure 2). In order to test our hypothesis (2) that the males in the group-housed condition (n = 8) would develop more accurate renditions of their tutors’ songs than males in the social isolation condition (n = 3), we first took the difference of the average SI scores at days 60 and 70 from the average SI scores at days 90 and 100 (MD = -0.029). A Mann-Whitney U test indicated that median SI scores across time bins did not significantly differ between group-housed (Mdn = -0.019, IQR = ± 0.045) and isolated males (Mdn =-0.019, IQR = ± 0.045; U = 10.000, p = .776, r = 0.123) which suggested that housing conditions had no effect on song accuracy relative to their tutors (Figure 3). Exploratory Analyses Given the unusually low SI scores of both conditions across all time bins (Mdn60 = 0.229, IQR = ± 0.180; Mdn70 = 0.232, IQR = ± 0.133; Mdn80 = 0.245, IQR = ± 0.112; Mdn90 = 0.251, IQR = ± 0.099; Mdn100 = 0.215, IQR = ± 0.192), we conducted exploratory analyses by comparing song accuracy relative to siblings to assess if fraternal inhibition (Tchernichovski & Nottebohm, 1998) occurred. SI scores calculations were conducted in all possible permutations, alternating a sibling as the tutor and pupil, while attempting to match siblings by age (Mage = 94.65; range: 80 dph - 101 dph). The resulting SI scores of each permutation calculated were then each treated as single observations and entered into SPSS. We repeated the process outlined above for the social isolation condition male pairs (Mage = 90; range: 70 dph - 100 dph) which left us with 23 total observations. Initial survey of the data showed us that the siblings’ songs 15 had converged and become more similar to one another than to their respective tutors. We then ran a Mann-Whitney U test, including all groups in the analysis, and determined that the jump in median SI scores relative to their siblings (n = 10; Mdn = 0.497, IQR = ± 0.247) was statistically significant from the median SI scores relative to their tutors (n = 9; Mdn = 0.276, IQR = ± 0.106; U = 21.00, p = .007, r = .115). To determine if there was a meaningful difference between conditions, we first compared the median SI scores relative to siblings between groups using a Mann-Whitney U test and found the median SI score of group-housed males (n = 10, Mdn = 0.546, IQR = ± 0.162) was significantly greater than the median score of isolated males (n = 4, Mdn = 0.252, IQR = ± 0.269; U = 1.00, p = .004, r = 0.267; Figure 4). Following up with the significant difference in median SI scores between grouphoused and isolated siblings, we further assessed whether the SI scores relative to the siblings of the group-housed condition were meaningfully higher than their own scores relative to their tutors (n = 19; range: 70 dph - 100 dph; M = 94.65 dph). We found that the median SI scores relative to siblings (n = 10, Mdn = 0.546, IQR = ± 0.162) was significantly greater than the median SI scores relative to their tutors (n = 9, Mdn =0.276, IQR = ± 0.111) suggesting that group-housed males developed more accurate copies of their siblings’ rather than their tutors’ songs (U = 1.000, p < .001, r = 0.189; Figure 5). 16 Figure 1. Spectrograms of a tutor with two of its pupils, both siblings, in the grouphousing and social isolation conditions. Note. The y-axis represents frequency (0-11 kHz) in thousands and the bar on the x-axis denotes 250-ms intervals. a, Spectrogram depicts the tutor’s motifs (Orange 923). b, Spectrograms of a pupil (Orange148) at 73 dph in the group-housing condition. c, Spectrogram of group-housing pupil (Orange48) at 95 dph. Visual inspection of the pupil’s spectrograms between 73 and 95 dph did not indicate qualitative change occurred in the song topography (spectrogram c) relative to their tutor (spectrogram a). d, Spectrogram of a pupil (Orange157) at 73 dph in the social-isolation condition. e, Spectrogram of Orange157 at 93 dph. Visual comparisons of Orange157’s spectrograms between 73 (spectrogram d) and 93 dph (spectrogram e) do suggest qualitative change occurred in either males’ song topography, and visual inspection shows that the pupil’s song is moFigure 1re accurate to the tutor than is suggested by the SI 17 scores, again showing that the SI scores do not seem to be accurately reflecting song accuracy (Orange923, spectrogram a). Figure 2: Median SI Scores Across Groups. Note. Median SI scores remained relatively stable for both groups (n = 11) across each time bin. 18 Figure 3: Median SI Scores Across Time Bins and Between Groups. Note. Sample sizes are delineated by the numbers labeled at each time point. 19 Figure 4: Median SI Scores Relative to Siblings Between Groups. Note. This graph demonstrates that the median SI score of group-housed males (n = 10, Mdn = 0.546, IQR = ± 0.162) was significantly greater than the median score of isolated males (n = 4, Mdn = 0.252, IQR = ± 0.269; U = 1.00, p = .004, r = 0.267), as we would expect given that socially isolated males did not have opportunities to interact with their siblings. 20 Figure 5: Comparison of Group-housed Condition Median SI Scores Relative to Tutor and Siblings. Note. As this graph demonstrates, group-housed males sang significantly more accurate renditions of their siblings’ songs (n = 10, Mdn = 0.546, IQR = ± 0.162) than that of their tutors’ (n = 9, Mdn =0.276, IQR = ± 0.111), suggesting that grouphoused males continued to learn and modify their songs in response to social experience (U = 1.000, p < .001, r = 0.189). 21 Discussion Our social lives were disrupted in 2020 as “shelter in place” commands swept the globe to limit transmission of a poorly understood virus (UNSD, 2020). Virologists are tirelessly learning about the novel virus, and psychologists, just as importantly, are working towards a greater understanding of our novel social environment: prolonged isolation (Kazak, 2020). The goal of this study was to explore whether zebra finches (Taeniopygia guttata) could be used as a model organism for research efforts articulating the consequences of long-term isolation on learning. The literature is clear on the beneficial and promotive qualities social interaction has on human development. Peer interactions promote language development and refinement along with stimulating cognitive growth and social skills (Bolhuis et al., 2010; Chib et al., 2018; Chugani et al., 2001; Coolahan et al., 2000; Fawcett et al., 2005; Fromkin et al., 1974). However, less is known regarding the consequences of social isolation. Therefore, psychologists should continue striving to understand the determinant role peer interactions and isolation have on human development. The goal of this study was to investigate how social isolation affects learning. We hypothesized that song accuracy would increase for both groups, but that birds in the social housing condition would have greater imitation scores relative to their tutor than birds in the isolation condition. Our hypotheses were not supported: we found that song accuracy for both groups did not improve, and the group-housing condition did not perform better than the isolation condition. This directly contrasts with past research which suggests song accuracy increases over time and that social interaction facilitates 22 song development (Caruso-Peck & Goldstein, 2019; Deregǹacourt & Gahr, 2013; Heston & White, 2015; Jesse & Riebel, 2012). In light of these unusual findings, we conducted exploratory analyses to measure the extent to which the songs of group-housed males converged or became more similar than their tutors’ songs. We found that group-housed males produced songs more similar to each other than their tutors. These findings directly conflict with prior literature on the facilitative effect social interactions have on song development (Carouso-Peck & Goldstein, 2019). Moreover, our results conflict with the assumption grounded in the findings of Immelmann (1969) which suggested that sensory learning ceases during motor practice between approximately 35 and 60 dph. Instead, our study supports the notion that song plasticity extends into adulthood (Jones et al., 1996; Price, 1979). The present study’s findings along with studies on extended plasticity (Jones et al, 1996; Price, 1979), suggest that social interactions are dynamic factors in a nested system that shape song development. Future researchers will have to identify the components of this system to fully understand the effects of social interactions on song development and researchers must continue to probe the extent to which song remains plastic into adulthood. Social Inhibition vs. Social Facilitation The social dynamics of male zebra finches are complex, involving multiple processes that can work in opposition to one another. Fraternal inhibition, spontaneous syllable production, and syllable convergence can lead males to develop poor copies of the tutor’s song (Jones et al., 1996; Tchernichovski & Nottebohm, 1998). Rather than 23 social processes functioning as protective or facilitative factors (Carouso-Peck & Goldstein, 2019; Deregǹacourt & Gahr, 2013), peer interactions might impair song development (Jones et al., 1996; Tchernichovski & Nottebohm, 1998). Jones et al. (1996), discovered that males in group housing spontaneously produce unique syllables, which converge over time, leading to deviation from the tutor song. It seems spontaneous syllable production potentially occurred in our grouphousing condition resulting in fraternal inhibition of song, possibly explaining the poor SI scores (Tchernichovski & Nottebohm, 1998). Future researchers should revisit the foundational studies on zebra finches (e.g., Immelmann, 1969; Price, 1979) and attempt to replicate these findings. Such an effort would require controlling for genetic differences by sampling brothers and probing the extent that song remains plastic across development by sampling across ages. Additionally, it is important to continue systematically interrogating the effect of social interactions on song development by using group-housed samples of differing sizes to assess if differences in sample size result in significant differences in song accuracy and syllable convergence. This will allow researchers to further understand if brothers serve as protective factors (Deregǹacourt & Gahr, 2013) or if housing zebra finches with conspecifics and siblings have more deleterious effects than previously understood. Ecological Explanations Aside from fraternal inhibition, Brainard and Mets (2019) found that if song instruction is specifically tailored to the genetic predispositions of slow singing birds, typically worse learners, they learn just as well as fast singing birds. Carouso-Peck and 24 Goldstein (2019) found that males who received feedback from a female produced songs more accurately to their tutor. Isolated birds had no females to attract, potentially disincentivizing them to copy their tutor’s song. Their song ability may have been properly encoded but perhaps not revealed. Furthermore, the group-housed subjects perhaps were introducing more variability into their songs to attract a female. Future research is needed to understand if the presence of a single female paired with a single male may elicit more accurate copies of the tutor song. Zebra finches in isolation, in past studies (e.g., Perez et al., 2012), have shown an increase in plasma corticosterone, a stress hormone, which may affect their vocal output such as having lower vocal activity and may result in the production of vocalizations that are longer and higher in pitch (Perez et al., 2012). Therefore, stress might be a confounding factor that led our isolated birds to produce erroneous songs. However, past research has shown increases in learning over time (Heston & White, 2015), while ours did not, which suggests a replication issue. This may be because Heston and White (2015) utilized 200 asymmetric pairwise comparisons of 20 renditions of the pupil’s typical motif with 10 renditions of the tutor motif, and our study utilized 25 pupil bouts compared to 5 tutor motifs per subject. They furthermore only reported the upper-third quartile score from these comparisons, so the percentage of the tutor song copied was not underestimated. We made no such compensations in our 25 study; the opposing finding indicates that future researchers need to continue exploring how song trajectories differ under various experimental manipulations. Ecological and Instrumentation Limitations Our study is limited in its ecological validity. Zebra finches are naturally gregarious, living and communicating in large flocks, and social isolation is rare. Furthermore, juveniles do not typically separate from their parents and continue to live in flocks throughout their lifespan (Templeton et al., 2014). Our small sample size does not allow for adequate external validity to humans, nor does it allow for individual differences between the birds to be averaged. Additionally, there may have been researcher bias in determining what was a bout and a motif in song analysis. There are instances when younger birds have immature songs making it difficult to discern a bout from a call. However, in attempts to mitigate this bias, each researcher analyzed all pupils of one tutor, while blinded to age and condition. Furthermore, the similarity-index software does not seem to accurately reflect the actual song accuracy displayed by the birds. In some instances, the software did seem to accurately reflect song accuracy, but there were also instances when that was not the case, making the internal validity of the similarity-index software questionable. These discrepancies may have been a result of researcher bias in choosing bouts and motifs, but the high interrater reliability suggests there may be a bigger issue with the software itself. Instrumentation effects may have been present as sounds separate from songs (e.g., noise from chamber fans, wing flapping, hopping, unlearned calls) may have been included in analysis, potentially affecting SI scores. Furthermore, only songs between 10 26 am and 12 pm were analyzed and in some cases, our subjects at certain recording intervals (i.e., 60 dph) had no songs within that time. Future research should broaden the time window to ensure adequate acquisition of song, allowing for a more comprehensive learning curve. Due to resource constraints, we were unable to consistently record grouphoused subjects in the same chambers, potentially leading to variations in recording quality and differences in overall acoustics. Implications and Future Directions Researchers must continue exploring the roles of syllables convergence and fraternal inhibition in zebra finch song development. While our research, along with the findings of Tchernichovski and Nottebohm (1998), does suggest fraternal inhibition can occur, the specific mechanisms for how it can occur are ambiguous. For example, our findings only suggest that conspecifics might cause spontaneous syllable production, but it remains unclear how many conspecifics are required for fraternal inhibition to occur. Furthermore, it is unclear if the timeline for spontaneous syllable production is the same for crystallization of a tutor’s song. Future research is required to answer these questions to understand the potential consequences of housing male juveniles together. Furthermore, researchers should experimentally test the effects of social isolation on learning with humans. Social interactions are correlated with increased attention levels and academic performance (Coolahan et al., 2000). However, there seem to be few experimental studies assessing this phenomenon due to the ethical issues surrounding human isolation. Fawcett and Garton (2005) suggest, however, that peer interaction in problem-solving tasks led to greater overall academic performance, suggesting 27 socialization aids in learning (Spengeman & Knauft, 2021). However, the findings of Hanushek et al. (2003) directly contrast with the facilitative effects of peer interactions suggested by Fawcett and Garton (2005). Further research will need to be conducted to develop a deeper understanding of the potential consequences of human isolation. Conclusion Overall, social isolation and its effects on learning are complicated, drawing forth little agreement. For the past year, humans have been in a state of unparalleled social isolation perhaps with many unknown consequences, with few solutions. The use of social animals in laboratory experiments allows researchers to potentially bridge this gap in understanding, as experiments can be more rigorously controlled. Researchers must continue to study how social isolation affects learning to develop interventions to mitigate the deleterious effect of isolation on cognitive development. Ideally, future research will continue to explore this complex issue in the hopes of proposing solutions to aid in the learning outcomes of students who are isolated from their peers. 28 Bibliography Bakar, J., & Orji, C. T. (2018). Effects of reciprocal peer tutoring and direct learning environment on sophomores’ academic achievement in electronic and computer fundamentals. Education and Information Technologies, 24(2), 1035–1055. Bölting, S., & von Engelhardt, N. (2017). Effects of the social environment during adolescence on the development of social behaviour, hormones and morphology in male zebra finches (Taeniopygia guttata). Frontiers in Zoology, 14(5). Bolhuis, J. J., Okanoya, K., & Scharff, C. (2010). Twitter evolution: Converging mechanisms in birdsong and human speech. Nature Reviews Neuroscience, 11(11), 747-759. doi:10.1038/nrn2931 Brainard, M. S., & Doupe, A. J. (2002). What songbirds teach us about learning. Nature, 417(6886), 351–358. Brainard, M. S., & Doupe, A. J. (2013). Translating birdsong: Songbirds as a model for basic and applied medical research. Annual Review of Neuroscience, 36(1), 489– 517. Carouso-Peck, S., & Goldstein, M. H. (2019). Female social feedback reveals non imitative mechanisms of vocal learning in zebra finches. Current Biology, 29(4), 631-636. 29 Coolahan, K., Fantuzzo, J., Mendez, J., & McDermott, P. (2000). Preschool peer Interactions and readiness to learn: Relationships between classroom peer play and learning behaviors and conduct. Journal of Educational Psychology, 92(3), 458–465. Chib, V. S., Adachi, R., & O’Doherty, J. P. (2018). Neural substrates of social facilitation effects on incentive-based performance. Social Cognitive and Affective Neuroscience, 13(4), 391–403. Chugani, H. T., Behen, M. E., Muzik, O., Juhász, C., Nagy, F., & Chugani, D. C. (2001). Local brain functional activity following early deprivation: A study of postinstitutionalized Romanian orphans. NeuroImage, 14(6), 1290–1301. https:// doi.org/10.1006/nimg.2001.0917 Deregnaucourt, S., & Gahr, M. (2013). Horizontal transmission of the father’s song in the zebra finch (Taeniopygia guttata). Biology Letters (2005), 9(4), 20130247– 20130247. Doupe, A. J., & Kuhl, P. K. (1999). Birdsong and human speech: Common themes and mechanisms. Annual Review of Neuroscience, 22(1), 567–631. Fawcett, L. M., & Garton, A. F. (2005). The effect of peer collaboration on children’s problem-solving ability. British Journal of Educational Psychology, 75(2), 157– 169. 30 Fromkin, V., Krashen, S., Curtiss, S., Rigler, D., & Rigler, M. (1974). The development of language in Genie: A case of language acquisition beyond the “critical period.” Brain and Language, 1(1), 81–107. Hanushek, E. A., Kain, J. F., Markman, J. M., & Rivkin, S. G. (2003). Does peer ability affect student achievement? Journal of Applied Econometrics, 18(5), 527–544. Heston, J. B., & White, S. A. (2015). Behavior-linked FoxP2 regulation enables zebra finch vocal learning. The Journal of Neuroscience, 35(7), 2885–2894. https:// doi.org/10.1523/jneurosci.3715-14.2015 Immelmann, K. (1969). Song development in the zebra finch and other estrildid finches. In R. A. Hinde (Ed.), Bird vocalizations: Their relation to current problems in biology and psychology: Essays presented to W. H. Thorpe. (pp. 61-74). New York, NY: Cambridge University Press. Jesse, F., & Riebel, K. (2012). Social facilitation of male song by male and female conspecifics in the zebra finch, Taeniopygia guttata. Behavioural Processes, 91(3), 262–266. Jones, A., Ten Cate, C., & Slater, P. (1996). Early experience and plasticity of song in adult male zebra finches (Taeniopygia guttata). Journal of Comparative Psychology, 110, 354–369. Kazak, A. E. (2020). Psychology is an essential science: American Psychologist highlights the role of psychology in understanding and addressing COVID-19. American Psychologist, 605–606. 31 Long, M. A., & Fee, M. S. (2008). Using temperature to analyse temporal dynamics in the songbird motor pathway. Nature, 456(7219), 189–194. Long, M. A., Katlowitz, K. A., Svirsky, M. A., Clary, R. C., Byun, T. M., Majaj, N., Oya, H., Howard, M. A., & Greenlee, J. D. W. (2016). Functional segregation of cortical regions underlying speech timing and articulation. Neuron, 89(6), 1187– 1193. Love, J. M., Kisker, E. E., Ross, C., Raikes, H., Constantine, J., Boller, K., Brooks-Gunn, J., Chazan-Cohen, R., Tarullo, L. B., Brady-Smith, C., Fuligni, A. S., Schochet, P. Z., Paulsell, D., & Vogel, C. (2005). The effectiveness of early head start for 3year-old children and their parents: Lessons for policy and programs. Developmental Psychology, 41(6), 885–901. Martins, P. T., & Boeckx, C. (2020). Vocal learning: Beyond the continuum. PLOS Biology, 18(3), 1-18. . Mandelblat-Cerf, Y., & Fee, M. S. (2014). An automated procedure for evaluating song imitation. PLOS ONE, 9(5), e96484. Mets, D. G., & Brainard, M. S. (2019). Learning is enhanced by tailoring instruction to individual genetic differences. ELife, 8. Perez, E. C., Elie, J. E., Soulage, C. O., Soula, H. A., Mathevon, N., & Vignal, C. (2012). The acoustic expression of stress in a songbird: Does corticosterone drive isolation-induced modifications of zebra finch calls? Hormones and Behavior, 61(4), 573–581. https://doi.org/10.1016/j.yhbeh.2012.02.004 32 Price, P. H. (1979). Developmental determinants of structure in zebra finch song. Journal of Comparative & Physiological Psychology, 93(2), 260–277. Slater, P. J. B., & Jones, A. E. (1998). Practice and song development in zebra finches. Behaviour, 135(8/9), 1125–1136. JSTOR. Spengeman, J.D., Knauft, M. (2021) “Reengagement:”A relationship-based intervention for disengaged students [Conference presentation April 13]. Whitman Undergraduate Conference 2021, Walla Walla, WA, United States. Tchernichovski, O., & Nottebohm, F. (1998). Social inhibition of song imitation among sibling male zebra finches. Proceedings of the National Academy of Sciences, 95(15), 8951–8956. Templeton, C. N., Laland, K. N., & Boogert, N. J. (2014). Does song complexity correlate with problem-solving performance in flocks of zebra finches? Animal Behaviour, 92, 63–71. Wafters, J. J., & Ginns, I. S. (n.d.). Peer assisted learning: Impact on self-efficacy and achievement. 15. United Nations Sustainable Development Group. (2020). Policy Brief: Education during COVID-19 and beyond [Government]. United Nations. 33