Everyone knows of “hey google” and “Alexa,” and the way these speech recognition devices have vastly transformed people’s way of life, but there are greater possibilities with such systems including improving the early education of children. Many children are behind in their reading and speech acquisition skills, a direct result of lack of exposure to vocabulary and oral skills. While other companies have created intelligent tutoring systems, they do not possess the ability to pick up on the acquisition of children. In order to determine the success of a young child interacting with a social robot, a series of tasks were completed and recorded with an instructor, child and the robot JIBO. These sessions were analyzed to quantify patterns in interactions such as boredom, excitement, and frustration. The data concluded that boredom and prompts from the instructor occurred most often, showing that young children struggle the most with staying focused during the long tasks. The more interjections from the instructor, the longer the sessions lasted since they kept the children on track, but it was found only 18.7% of the recordings were a child speaking. It’s suggested the tasks be shortened so that there can be more success in the students completing all of the tasks to be recorded and analyzed for future sessions. Social robots can be used in classrooms, clinician offices and even at home to develop learning at a young age that is most beneficial for the future.

Demand for rechargeable electronics has increased reliance on lithium (Li) metal batteries. Unfortunately, repeated plating and stripping of Li metal results in dendrite formation, limiting safe and reliable use of Li as an electrode. Our project seeks to develop a free standing monolithic ionogel separator, which would function as an electrolyte while suppressing dendrite formation for long term cycling. We synthesized ionogel materials consisting of MTMOS, TMOS, VTMOS, formic acid, and LiFSi in ionic liquid using typical ionogel synthesis procedures. Galvanostatic measurements in coincells revealed consistent plating and stripping for up to 250 hours (for spin coated Cu) and up to 30 hours (for infiltrated celgard) without evidence of Li dendrite formation or short-circuiting. These results represent a “proof-of-concept” for the production of free-standing pseudo-solid electrolytes and suggests promising results for Li dendrite suppression.

Implementing efficient renewable energy sources is necessary to combat climate change. Existing models estimate net generation from rooftop solar installations utilizing single-junction cells (which use one material to absorb and convert light) versus more expensive multijunction cells (which use multiple materials and convert sunlight more efficiently). Further, power production estimations are often based on efficiency, representing the power converted for a standard spectrum (AM1.5G) without accounting for temporal changes. Spectral changes due to climate and season have significant impacts on the generation capability of solar cells, but data constraints and geographic variations hamper effective modeling. We hypothesize that a multijunction cell design employing a low-cost flat concentrator will generate significantly more energy from rooftop installations than current estimations of efficiency improvements. We applied Warmann’s* spectral binning method and sunlight data from Los Angeles to detailed balance calculations with realistic concentrator and cell properties to predict the cell’s power output. Which, we applied to the LA Solar Map to calculate LA’s total annual solar energy generation. We found that our dual-junction solar cell design is 39.48% efficient while the record silicon cell is 29.76% efficient under the AM1.5G spectrum. However, accounting for annual spectral changes, our design increases yearly energy produced to 67.56 kWh/m² compared to the Si-single junction cell, demonstrating the importance of comparing designs through annual energy produced. A 58% increase in energy production warrants considering multijunction cells for rooftop use because the significantly increased energy production better offsets higher manufacturing costs.

Currently, our inability to effectively edit the mitochondrial genome limits the development of metabolic disease models that can be used to test and create novel therapies for mitochondrial DNA (mtDNA) disorders. Mitochondrial myopathy, encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS) is a primary mitochondrial disease. The generation of MELAS-centered induced pluripotent stem cells (iPSCs) could be essential for developing metabolic disease models for future therapeutic research. Earlier studies have created iPSCs from primary cells with high-level mutant mtDNA and subsequently differentiated them into all three germ lines (Hamalainen et al, 2013). However, it is in our interest to create these iPSCs from cells that have been engineered with the mutation using the force-driven MitoPunch apparatus (Sercel et al, 2021), which transfers isolated mitochondria from donor cells into mitochondria-depleted recipient cells.Previous studies in the lab show successful iPSC reprogramming and differentiation of MitoPunch-engineered cells possessing wildtype mitochondria but did not observe mutant-containing iPSCs (Patananan et al, 2020). This could be due to 1) epigenetic alterations resulting from 2’,3’-dideoxycytidine (ddc) exposure during the mtDNA depletion process, 2) the presence of high mtDNA mutation levels creating a bioenergetic limitation, or 3) a mismatch between nDNA and transferred mtDNA, hampering mitochondrial-nuclear communication and coordination. Using two cell lines, a neonatal dermal fibroblast (NDF) and a MELAS line, we performed four combinatorial mitochondrial transfers, to test for the nDNA-mtDNA matching and the detrimental effects of mtDNA depletion. Successful reprogramming will generate mutant iPSCs that can be used in additional studies for disease modeling.

There is broad evidence that humans exhibit daily rhythms in cognitive performance, with peak achievement occurring during the day. This study will test the hypothesis that performance in verbal learning-- specifically in second language acquisition, varies in daily rhythm with peak performance in the late morning. We will take advantage of a newly developed virtual reality (VR) task using special ceramic goggles to carry out this work. Task wise, we test the ability of individuals to learn a new language (Spanish). We will teach users through conversations with an avatar, Antonio, in our immersive virtual reality environment. Afterward, we will test retention and recall through a conversation with a native speaker. Before and after two weeks of training, the test will be performed while a subject’s brain activity is monitored with functional magnetic resonance imaging (fMRI). We will test the hypothesis that subjects learn words and phrases better in the late morning and perform worse at night. Individuals trained and tested one of 6 times during the day and night (6:00, 10:00, 14:00, 18:00, 22:00). We expect to see improved performance at 10:00 and worse performances at 22:00. We expect greater synchrony in fMRI in brain regions involved in verbal learning and conversation skills at the time of peak performance, and substantial cortical thickness in the hippocampus after two months. Defining optimal times to learn languages has practical implications for schools, and professionals, and future work will determine if daily rhythms are under the control of the circadian system.