We often wonder what kind of information we need in multimedia materials. This presentation deals with the question using examples from gloss studies. Studies have examined what kind of glosses are helpful for learners to pick up words as they read a text with glosses. Studies have shown that glosses with text and pictures are more effective than text-only (Akbulut, 2007; Ramezanali & Faez, 2019). The dual coding theory (Sadoski & Paivio, 2001) explains the phenomenon: we learn something better when it is coded both verbally (text) and non-verbally (pictures). On the other hand, other studies have indicated that text-only glosses are better than or equal to text and pictures (Acha, 2009; Boers et al., 2017). This is based on the split attention hypothesis (Mayer & Moreno, 1998). When the information is presented through one channel, such as through the eyes, attention is split between text and picture. As a result, learners get distracted by the two types. Text-only produces better results since learners can focus on just one type. The presenter wishes to open the discussion on how to dissolve the discrepancy between the two theories and hopes to bring a better understanding of multimedia uses.