Explanation is an essential aim of science, and the dominant approach to scientific explanation is causal. According to causal accounts, scientific theories explain phenomena by appealing to causes or causal mechanisms. Wesley Salmon and James Woodward defend a causal approach to scientific explanation in the two main but different causal approaches in the literature. These two approaches clearly introduce and defend ontological commitments in a scientific theory of explanation. However, since the mid-2000s, there has been a defence of non-causal approaches that we call the new generation of defenders of non-causal explanations (compared to Hempel, Kitcher and their followers). This new generation introduces examples in different scientific fields like physics, mathematics, biology, and neuroscience, where causal explanations do not work. They argue for the need for non-causal explanations in these cases. However, this new generation is silent about the ontological implications of non-causal explanations. We argue that even these non-causal approaches have ontological commitments to weakly emergent properties and entities. To defend non-causal approaches to explanation, this new generation appeals to the role of scientific models in scientific explanation. They follow Ronald Giere's approach to scientific models, according to which scientific theories represent the target systems through scientific models, not mere physical laws. Some philosophers, Like Alisa Bokulich, go further than Giere and argue that, in some important cases, "fictional models" have explanatory power. She refers to semiclassical mechanics to defend these fictional models' explanatory power. We agree with the unavoidable role of non-causal accounts in scientific explanations. Also, we agree with Bokulich to defend the explanatory power of these highly idealized models in science. However, contrary to Bokulich, we argue that these semiclassical models include weakly emergent properties that give them explanatory power. To this aim, we introduce and defend weak emergentism and its two main characteristics. Then we show that these highly idealized models satisfy these two conditions and have weakly emergent properties. Finally, we use this approach in three more examples (damped oscillators, universal macro-behaviours, and rainbows) to show that even non-causal approaches to explanation have ontological commitments to weakly emergent properties.