United States

Rationale: 
Our pilot study has shown that electrical stimulation of specific brain regions through responsive neurostimulation (RNS) can improve memory consolidation [1]. This study aims to reinforce and investigate whether responsive neurostimulation (RNS) devices can improve memory function through the administration of verbal and non-verbal learning tasks. We hypothesize that direct cortical stimulation with RNS could improve memory. 

Methods:
In an outpatient setting, a verbal learning task (Hopkins Verbal Learning Test-Revised) was administered to patients with RNS implants in various brain regions (hippocampal, temporal, and insular) [2]. For the verbal task, patients read a list of 12 semantically related words and tried to recall the list in the 3 different trials, along with a delayed recall and recognition trial performed 20 minutes later. True or sham stimulations were administered to patients through Neuropace.

Results:
Most of the 34 participants were right-handed (88.2%) with the remainder being left-handed (11.8%). Of the two RNS electrodes, 44.1% had the first electrode in the hippocampus. Only 28 patients had RNS data collected. A paired two-sided t-test with an alpha level of .05 was performed for all statistical tests. Although current results show no statistical significance, there still is a trend for patients who underwent electrical stimulation at the time of learning to perform better on recall than those who had sham stimulations, especially in the delayed recall trial. ​

Conclusion:
The pilot study found that RNS stimulation can improve memory encoding; however, the smaller sample size prompted continuing investigation and analysis of RNS impact on memory. The current results show a trend for improvement of memory with electrical stimulation during verbal tasks. We plan to acquire more data to more effectively interpret the effects of RNS on memory. 


AMA Research Challenge 2023

The Impact of Brain Stimulation on Verbal Memory

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

Registration Required

ACL 2026

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

poster

Welcome to the **largest national, multi-specialty research** event featuring research from nearly 700 medical students, residents, fellows, and international medical graduates.

Use the buttons below or navigation menu to access the event.
 


[![](https://assets.underline.io/uploads/markdown_image/1/image/3e33a8c754cc49610b30950617d813e7.png)](https://underline.io/events/434/sessions?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/551404d429c4e274f9a54cbdd7081918.png)](https://underline.io/events/434/posters?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/677d92f566b880cebd36905fc738fc10.png)](https://underline.io/events/434/lounge)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/c7392833c7261c3874c5b43fa2f4df19.png)](https://underline.io/events/434/help-desk)
 

#### Event Details

The 2023 American Medical Association Research Challenge includes a poster symposium hall and a semifinals hall with the top scored research. **Log in at any time throughout the event to:**

* **View posters and presentations** on a variety of topics and specialties
* **Score posters in the semifinals** to help decide the top five finalists (AMA member exclusive opportunity)
* **Connect with participants** and provide feedback to help advance their research
* **Attend education sessions** with physician experts on key research topics

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge finals, present their research to an elite panel of judges, and compete to win a $10,000 grand prize sponsored by Laurel Road. 

### Grand Prize Sponsor

AMA would like to thank Laurel Road for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road is not involved in the judging or scoring process to select the winner. The winner will be announced on February 4, 2024 during the AMA Research Challenge finals. Payment of the grand prize will be directly from Laurel Road. See the AMA Research Challenge Official Rules for terms and conditions.

![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)

**There is no registration fee for this event! **

Register to explore research in a variety of topics. AMA members get exclusive access to score posters in the semifinals.

**Please use this link [here](https://www.ama-assn.org/about/events/ama-research-challenge) to register for free.**

Registration Is Required

Join the AMA Research Challenge poster symposium and semifinals to explore top research posters, score posters in the semifinals (an AMA member exclusive opportunity) and attend education sessions with experienced physicians to learn tips to advance your research and career. 

Background/ Objectives: 
Migraines are the second leading cause of disability worldwide. However, there is a paucity of research focused on the clinical presentation of chronic migraines (CM) in Native Hawaiians and other Pacific Islanders (NHOPI). This study aims to identify possible differences in the clinical presentation of CM in NHOPI in contrast to other Ethnocultural racial groups in Hawaii.
Methods:
A retrospective case-control study was conducted to examine the relationship between demographics, comorbidities, and chronic migraines. Data from a single neurological care center in Hawaii were used to identify adults aged ≥18 years diagnosed with CM between 2018-2023. Adults with CM were matched for age and sex to 3 controls (patients with radiculopathy who did not present with CM). 
Results:
Overall, 309 CM patients and 964 matched controls were identified for analysis. NHOPI populations accounted for 24.9% of CM and 26% of radiculopathy patients. NHOPIs, on average, were diagnosed with CM or radiculopathy 2-3 years earlier (p=0.07) and had a higher average BMI than their White counterparts (p<0.01). Among NHOPIs, CM cases were more likely to have public insurance coverage (p=0.04), hyperlipidemia (p<0.01), diabetes (p=0.02), and insomnia (p=0.01). They were also more likely to be on more medications than White patients (p<0.001). 
Conclusion:
NHOPI CM patients were more likely to be diagnosed earlier than White patients and had higher BMIs, as well as higher rates of hyperlipidemia, insomnia, diabetes, and polypharmacy. These findings have important implications for understanding underlying comorbidities among NHOPI CM patients and developing targeted interventions.

Chronic Migraine and Comorbidity Characterization: A Focus on Native Hawaiians and Other Pacific Islanders


Background: Rett Syndrome (RTT) is a progressive X-linked neurodevelopmental disorder affecting 1 in 10,000 live female births, caused by deletions in the methyl-CpG binding protein 2 gene (MECP2). It is characterized by developmental regression, stereotypical hand movements, motor dysfunction, language impairments and seizures. Retrospective case reports have indicated atypical pain perception in patients with RTT. Yet, studies in patients and the MeCP2-/+ heterozygous (Het) mouse model of RTT have been inconclusive, with both hypoalgesia and hyperalgesia reported. Since the MeCP2 protein is a critical transcriptional repressor factor expressed in sensory neurons, mechanisms have been speculated for the modulation of pain by MeCP2. However, few studies have explored its role in atypical pain perception with a translational model of RTT. A better characterization of abnormal pain behavior in an RTT model is critical for future mechanistic studies to provide better care for these patients. 

Methods: MeCP2-/+ female mice and age-matched wild-type (WT) female littermates were used. Two different ages were evaluated: 6-8 weeks (pre-symptomatic) and 4 months (symptomatic) when the RTT phenotype was present and consistent. Phenotype was assessed using a neurobehavioral score (looking at paw clench, respiratory pattern, and mobility) along with Catwalk Gait analysis. Pain sensitivity and thresholds were assessed to mechanical (Von Frey test) and heat (Hargreaves test) stimuli.

Results: MeCP2-/+ mice showed variation in their sensitivity to pain as their disease progressed. Pre-symptomatic mice were hypersensitive to both mechanical and thermal pain, shown by increased responses to tactile stimulus on Von Frey filament testing and decreased paw withdrawal latency on thermal Hargreaves testing. However, with the progression of disease phenotype, MeCP2-/+ mice became hyposensitive to both mechanical and thermal pain. These changes appeared to be independent of their motor ability.

Conclusion: MeCP2-/+ female mice represent a useful preclinical RTT model and exhibit abnormal pain behavior, with initial hyperalgesia and eventual hypoalgesia that appears to be independent of motor function. Since MeCP2 overexpression has been associated with increased pain tolerance, lack of MeCP2 in the Het female mice may contribute to the early pain hypersensitivity. Future studies are needed to explore the cellular and molecular mechanisms involved in this altered pain sensation during the progression of RTT and for developing stage-dependent therapeutic targets.

Changes in pain sensitivity with the progression of Rett Syndrome in MeCP2-/+ female mice

Introduction: Diagnostic criteria for Alzheimer’s Disease (AD) and Frontotemporal Dementia (FTD) are clinical, and determining specific etiology can be difficult even for experienced clinicians. Despite this, it is more important than ever to distinguish the two given the recent advent of novel therapies targeting disease-specific pathophysiologic mechanisms. Since clinical presentation can be nuanced, healthcare providers are increasingly turning to imaging and other biomarkers to aid in diagnosis of FTD and AD, and automated software tools like Neuroreader (NR) have been developed to provide objective, reliable volumetric data from brain imaging. Further, volumetric indices specifically between anterior and posterior brain areas have proved useful in distinguishing types of dementia in research cohorts. However, existing indices can be complicated to use and require further validation in the clinical setting. The objective of this study is to provide initial validation for the use of a simplified, easy-to-use anterior-posterior index (API) from NR data in distinguishing FTD and AD in a clinical cohort.
Methods: A retrospective chart review was completed to gather demographics, memory diagnoses over time, and cognitive screening and NR data in patients at a large academic medical center with a diagnosis of FTD or AD.
We then derived a simplified API to reflect NR data: API=((logV_A/V_P)⁡-μ)/σ where V_A is weighted volume of frontal and temporal lobes and V_P of parietal and occipital lobes derived from NR data. μ and σ are the mean and standard deviation of logV_A/V_P computed for AD patients only. Receiver operating curves (ROC) and 2 separate regression analyses (API and the parietal lobe) assessed the efficacy of the API vs. single brain areas in predicting the diagnosis of AD vs FTD as it was determined clinically using current diagnostic criteria.
Results: 39 patients with FTD and 78 patients with AD were included. The samples are similar in terms of demographics, cognitive testing score, and MRI type. The API had an excellent performance in distinguishing AD from FTD with an area under the curve (AUC) of 0.82 in addition to a positive association with diagnostic classification on logistic regression analysis (B=1.491, p<0.001). Most brain areas in isolation performed poorly, with the exception of the parietal lobe which had an AUC of 0.70, indicating an acceptable performance.
Conclusion: The API biomarker can successfully distinguish AD and FTD with an excellent level of performance, performing better than any studied brain area alone. The results of our study validate the API in a clinical setting and indicate that it could be a useful diagnostic tool for clinicians when distinguishing AD from FTD.

The Use of an Anterior-Posterior Atrophy Index to Distinguish Alzheimer’s Disease from Frontotemporal Disorders: An Automated Volumetric MRI Study

Background Patient-proxies often provide informed consent for neurocritical care research. Informed consent success depends on research staff availability, rapport-building time between patients-proxy, language barriers, health literacy, cultural biases that affect patient comprehension, proxy's distress with acute brain injury news, and perceived patient choices. Pandemic-related visitor restrictions made proxy consents harder to obtain. Two clinical trials highlight the importance of remote e-consent and proposed solutions.
Methods Consent workflow description for Study A utilizing multiple modality e-consent with remote investigators compared to Study B using in-person dedicated staff when remote e-consent was not allowed.
Results 
Study A contacted 51 of 70 eligible patients. Only one consent was obtained in-person during the 3-month recruitment period when IRB only allowed in-person consents. After remote e-consent approval, 50 patient proxies were mailed consent to sign after patient discharge, taking 6 hours of remote-investigator time. Only one returned signed form after 21 days. 49 proxies were phoned by Doximity over 6 months, taking 3.5 min/eligible subject with one redial every month (average 4 calls/subject), with 17 rejected, 11 unreachable. 8/21 phone proxies received DocuSign®-based consent emails and returned signed forms in average 36 days (4 hrs-144 days). Mail yielded 2% and email-based DocuSign consent 38%, compared to 1% in-person. 
Study B: Of 81 eligible over 6 months, 50 consented in-person or by-mail through 3 in-house investigators with a 62% success rate, with average of 48 minutes/consent (15–50 minutes), and 2 consents received by-mail. Getting consent in-person while patient was hospitalized, meeting proxies in parking lots when visiting premises for unrelated purposes, and dedicated unit staff increased success rate.
Conclusion 
Email-based DocuSign® e-consent through remote investigators can overcome pandemic-induced visitor limitations. Our institutional Digital Communications Governance Committee is exploring HIPAA-compliant text-based messaging solutions using cloud-based communication platform (Twilio®) to facilitate remote consenting access.


Pandemic-Induced Challenges in Consenting Patients andProposed Solutions for Neurocritical Care Trials

Background
Traumatic brain injury (TBI) is a leading cause of mortality and morbidity worldwide. Pupillometry is a noninvasive method to assess for preserved midbrain function. Specifically, poor pupil reactivity has been shown to be an early predictor of neurological deterioration and can signify impending herniation hours before clinical symptoms arise. In patients with TBI, absent pupil reactivity can stratify severity and prognosticate outcome. However, abnormal pupil phenotypes over the first 72 hours in mild, moderate, and severe TBI patients is poorly characterized. We calculated the incidence and burden of abnormal pupil phenotypes and investigated their relationship to unfavorable patient outcome at discharge (death or disposition to hospice or long-term care).
Methods
We performed a single-center retrospective observational study of 137 TBI patients with quantitative pupil measurements using the Neuroptics pupillometer. We calculated the median Neurologic Pupil Index (a metric of pupil reactivity ranging from 0-5) over 72 hours and compared cumulative burdens of abnormal pupil phenotypes (unilateral poor pupil reactivity and bilateral poor pupil reactivity) for mild, moderate, and severe TBI patients using ANOVA tests. We then used logistic regression to test the association of median average NPi and bilateral poor pupil reactivity burden with death or disposition to hospice or long-term care adjusting for age, sex, black race, TBI severity, and penetrating v. blunt injury. 
Results
Of 137 patients, 97 were male and median age was 59 [25th percentile 38, 75th percentile 72], of whom 49 (36%) had unfavorable outcomes. Fifty patients had mild TBI, 21 had moderate and 66 had severe TBI. Unilateral poor pupil reactivity occurred in 12 (24%), 6 (29%) and 32 (48%) patients with mild, moderate, and severe TBI. Bilateral poor pupil reactivity occurred in 3 (6%), 2 (10%), and 19 (29%) patients with mild, moderate, and severe TBI. After adjusting for confounders, median NPi over 72 hours was inversely associated with unfavorable outcome (0.22 [0.07-0.51]; p=0.004), while burden of bilateral poor pupil reactivity was associated with unfavorable outcomes (1.23 [1.08-1.45]; p=0.004).
Conclusion
Lower median NPi and higher burden of bilateral poor pupil reactivity over 72 hours in TBI patients is associated with unfavorable outcomes in TBI patients.

Prognostic Pupil Reactivity Markers within the first 72 hours of Traumatic Brain Injury

Amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig disease, is a fatal neurodegenerative disease affecting motor neurons in the brain and spinal cord. Progressive muscle weakness, atrophy, and spasticity characterize the condition, which eventually leads to paralysis and respiratory failure. There is currently no cure for ALS, and the standard of care is supportive, with riluzole being one of the few FDA approved medications that has been shown to slightly slow disease progression along with Radicava (edaravone) and RELYVRIO (AMX0035) approved in 2022 with Qalsody (tofersen) being approved in 2023. With limited options for medications, the use of mesenchymal stem cells (MSCs) in the treatment of ALS is a new area of research in regenerative medicine. MSCs are multipotent stem cells that can differentiate into a variety of cell types, including neurons, and have immunomodulatory and anti-inflammatory properties, making them an appealing therapy for neurodegenerative diseases such as ALS. This review examines the safety and efficacy of MSC therapy for ALS in three Phase II randomized controlled trials. The trials involved administering MSCs via various routes, including intrathecal and intravenous, and assessing outcomes such as survival, disease progression, and quality of life. 


Mesenchymal Stem Cell Therapy for Amyotrophic Lateral Sclerosis: A Narrative Review


Background 
Benefits of early skin-to-skin contact (SSC) are widely documented, including improved breastfeeding outcomes. While promoting immediate SSC is standard practice for vaginal birth, it happens less often after cesarean birth.  It is not known how changes in hospital practices and staff shortages during the COVID-19 pandemic have influenced the practice of SSC in the operating room.  This study aims to identify the relationship between SSC after cesarean birth and breastfeeding and compare SSC rates before and during the COVID-19 pandemic. 
Methods 
This was a retrospective, cohort chart review. Sample was 244 subjects who had scheduled cesarean deliveries at term during April-October 2019 and April-October 2020. Primary outcome was exclusive breastfeeding on discharge. Secondary outcomes were time to initiate breastfeeding, exclusive breastfeeding at postpartum visit and rate of SSC in the OR in 2019 compared to 2020. 
Results 
SSC in the first 3 days after birth was associated with feeding type on discharge (p<0.001) and/or at the postpartum visit (p=0.04). The study did not identify a statistically significant association between location of SSC and feeding type either on discharge (p=0.55) or at the postpartum visit (p=0.55). Infants who had SSC in the OR had significantly earlier initiation of breastfeeding (p<0.001). There was an increase in SSC in the OR between 2019 and 2020 (27% and 39%, respectively).  
Conclusion 
SSC practices after cesarean birth are associated with breastfeeding outcomes. Knowledge of this relationship will be important for optimizing surrounding SSC practices and breastfeeding support in the future.

Skin-to-skin contact and breastfeeding after cesarean birth at UIHC, before and during the COVID-19 pandemic

Background: Sonographic measurement of foetal biparietal diameter is a
well-accepted predictor of gestational age. However, there is a high variability
in the calculated gestational age which increases as pregnancy progresses with a
maximum difference approximating 3.6 weeks in the third trimester. Conditions
like oligohydramnios can alter the shape of the foetal skull, affect the BPD and
increase variability. The present study is being undertaken to measure the
transverse cerebellar diameter (TCD) to validate it as an additional
morphological measurement of foetal growth with less variability.
The cerebellum and posterior fossa are aligned perpendicular to the plane of
maximum extrinsic compression. Hence they can withstand deformation by
extrinsic pressure than the parietal bones and can be a more accurate parameter
for determining gestational age.
Methods: A cross-sectional study was done in 100 healthy women with
uncomplicated pregnancy between the 28th week of gestation and 40th week of
gestation over three years in outpatient and in-patient sections of Obstetrics and
Gynaecology Department of SRM Hospital and Research Institute,
Kattankulathur, Chennai. This study was undertaken to determine a correlation
between the transverse cerebellar diameter and the gestational age as
determined by the last menstrual period and other sonographic parameters like
biparietal diameter and femur length. After satisfying the inclusion and
exclusion criteria, Transabdominal ultrasonography was performed with the
patient in a supine position. The Pearson Correlation technique was used to test
the direction and strength of the relationship between gestational age obtained
using transverse cerebellar diameter vs individual parameters such as biparietal
diameter and femoral length.
Results: A scatter diagram showing the correlation and regression analysis of
the Gestational age with BPD in the third trimester. Here TCD is correlated with
BPDby87.1 %. A scatter diagram showing the correlation and regression
analysis of the TCD with FL in the third trimester. Here TCD is correlated with
FL by 93.0 %
Conclusion: Obstetric sonography was performed in 100 pregnant women with
uncomplicated pregnancy with known LMP, previous history of normal
menstrual cycle, and no exclusion criteria were included in the study. Foetal
biometry evaluated includes BPD, FL, and TCD. The nomogram of the TCD
shows that there is a linear relationship between cerebellar growth and
gestational age. So TCD can be used as a reliable parameter for determining
gestational age

 Accuracy of transcerebellar diameter in estimation of gestational age in third trimester and its comparison with conventional parameters such as biparietal diameter and femur length

Introduction
Trauma complicates 1 in 12 pregnancies and is the leading cause of mortality outside of obstetric-related complications. Guidelines for trauma activation in pregnancy differ from non-pregnant populations but are not universally established. This study aims to investigate under or overtriage of our pregnant patient population according to our established trauma triage criteria.
Methods
A retrospective study including all patients with pregnancy and trauma ICD-10 codes between January 2017-December 2020 at a level 1 trauma center was included. Patient demographics, vitals, injuries, service involvement during admission, gestational age, and obstetric and medical history were collected and compared to our hospital’s trauma triage criteria to evaluate undertriage in this patient population.
Results
330 patients were identified based on inclusion criteria. Trauma evaluation only occurred in 13.0% of patients and 30.9% of patients received obstetric (OB) evaluation. According to our institutional criteria, 55.8% of all pregnant patients had a missed trauma evaluation. 87.2% (n=130/149) of patients ≥20 weeks had a missed trauma evaluation (Fig. 1). For patients presenting directly to the trauma bay, patients ≥20 weeks received OB evaluation 100% (n=11) of the time, and patients <20 weeks received OB evaluation only 53.9% (n=7/13) of the time.
Conclusion
Pregnant patients are a vulnerable population, and thus, it is essential to triage these patients based on OB-specific criteria. This study has revealed that most OB patients are undertriaged according to our institution's criteria, prompting further investigation to understand barriers to appropriate triage.

Trauma Trends and Outcomes in Obstetrical Patients at a Level 1 Trauma Center

Background
Pregnant women were one of the most vulnerable populations during the Coronavirus Disease 2019 (COVID-19) pandemic. Understanding how the pandemic influenced prenatal care utilization and birth outcomes of pregnant populations is essential to guiding disaster preparedness in the face of future global threats. We sought to evaluate the COVID-19 pandemic’s influence on the adequacy of prenatal care and birth outcomes.

Methods
Retrospective cohort study compared prenatal care adequacy, diagnostic testing, and gestational age of 481 patients seen at an urban medical center in New York City during the pandemic (October 1, 2020 and October 14, 2020), and in the corresponding pre-pandemic period (October 1, 2019 and October 14, 2019) using student t-test, Mann-Whitney, and chi-squared tests.

Results
Patients seen during the pandemic were more likely to receive Adequate Prenatal Care (OR 2.634 95%CI 1.672-4.184 p < 0.001) than pre-pandemic patients (OR 0.385 95% CI 0.244-0.606 p < 0.001). Adequate Prenatal Care (OR 0.102 95%CI 0.014-0.763 p = 0.005) was negatively associated with early preterm pregnancy (32-36 weeks). There were no statistically significant differences in gestational age between the pre-pandemic and pandemic cohorts (Moderate Preterm 32-36 weeks OR 1.145 95%CI 0.532-2.465 p=0.845, Early Term 37-38 weeks OR 1.049 95%CI 0.658-1.671 p=0.905, Full Term OR 1.044 95% CI 0.689-1.581 p=0.916). 

Conclusion
Though the pandemic imposed unprecedented stresses upon expectant mothers, patients had significantly greater likelihoods of receiving adequate prenatal care during the pandemic, which may be attributed to the expansion of telehealth services and increased health awareness. Despite this increase in prenatal care adequacy, the pandemic had no statistically significant influence on preterm birth rates.

Premium content

Downloads

Next from AMA Research Challenge 2023

Chronic Migraine and Comorbidity Characterization: A Focus on Native Hawaiians and Other Pacific Islanders

Similar lecture

Unveiling the Limits of Large Language Models in Inferring Pragmatic Meaning from Non-Verbal Responses