Questioning the functional roles of visual representations in processes of scientific reasoning allows for different perspectives on the topic. On the one hand, we can focus our inquiry on the type of visual representation. Are we dealing with naturalistic images such as photographs, images that only appear to be naturalistic such as fMR images, a problem that Adina L. Roskies has made us aware of, or rather artificial ones, such as diagrams or maps, that seem to be closer to language than to images? Focussing on different types also means acknowledging their different capacities to transmit information and support reasoning processes. On the other hand, we can also call attention to the different kinds of epistemic desiderata we are interested in and what kind of contribution visual representations can make in these contexts. The most common question discussed in this regard, for instance by Laura Perini, is whether images can transmit propositional knowledge, that is, knowledge-that. Yet we might also wonder whether visual representations can help to transmit other kinds of knowledge such as knowing-how or knowledge by acquaintance as discussed by Bertrand Russell. Finally, we might analyse the capacity of visual representations to enable and facilitate understanding of certain phenomena. This last epistemic desideratum is particularly interesting as some philosophers, for instance Jonathan L. Kvanvig, think that the connection between understanding and knowledge, which is usually regarded as a very close one, should be considered anew. The common idea that knowing that p also implies understanding that p is called into question here. Contrary to this, it is argued that understanding adds a particular dimension to the epistemic project that is not necessarily part of the acquisition of knowledge. Here, an analogy can be made between the former case and the so-called knowledge-action-gap discussed in the debate about knowledge-how. An example might illustrate this point: apparently, people are able to acquire theoretical knowledge about how to play a musical instrument without being able to perform it. In the same way, people can acquire propositional knowledge, for instance via learning facts by rote, without understanding the phenomenon in the background. That the process of learning is incomplete in such a way often becomes apparent when people are unable to explain in their own words the content of what they hitherto have learnt. Another indicator of a lack of understanding shows up, if students are not able to apply newly acquired knowledge to solve problems which are not identical, but sufficiently similar to the one they got their knowledge from. Regarding things this way, that is, assuming that understanding can be regarded as an epistemic desideratum in its own right, makes it possible to analyse contributions of visual representations in this realm without devaluing their epistemic potential due to their working as heuristic tools only. Henk W. de Regt has emphasised this tool-character of scientific images. He argues that visual representations can be particularly useful to help scientists to understand certain phenomena in their area of research. A famous example discussed by him are Feynman diagrams. However, de Regt also thinks that it is due to a particular training of students that these tools can facilitate scientific understanding. Scientists are dependent on prior knowledge and skills to interpret images correctly, hence, it would be plainly wrong to regard scientific images as indispensable tools to facilitate scientific understanding. Empirical studies pursued by educational psychologists seem to support de Regt’s thesis. They point out the relevance of learners’ characteristics as an important criterion in decision processes on methods and tools to use in studying and teaching. Students’ background knowledge as well as their cognitive preferences play important roles when it comes to interpreting visually or verbally presented content. In this talk, the role of visual representations in facilitating scientific understanding will be discussed anew by starting from the assumption that, on the one hand, understanding can indeed be regarded as an epistemic desideratum in its own right and, on the other, by pointing out in what sense images can, despite the sceptical points just mentioned, play an indispensable role in this context. Not only will considerations of educational psychologists be taken into account, but also some particularities about the cognitive content of visual representations will be discussed to support this latter thesis. When it comes to epistemological considerations about the capacities of visual representations in scientific reasoning processes, the close relation between the functional and the typological level of scientific images will be analysed.