Visual signals represent one of the most common forms of communication in animals, and they play an important role in predator and mate recognition in many species. Visual signals can be expressed in terms of color, brightness, context, magnitude, ultraviolet (UV) reflectance, as well as fluorescence, and often times animal visual systems are specialized to discriminate between particular visual features that are more biologically relevant. My dissertation work examines these aspects using Heliconius butterflies. First I tested whether Heliconius erato communal roosting behavior confers a selective benefit in terms of predator avoidance through collective warning signaling, and explored how roost size influences predator behavior (Chapter 1). I found that magnitude, by means of roost size, plays a role in the effectiveness of roosts as anti-predator signals. Following roosting studies, I identified the relative importance of color versus pattern signals for both predator avoidance and mate attraction by these butterflies (Chapter 2). My results showed that color appears to be a more effective repel signal than pattern, and the same signals selected for by predators are most preferred by mates. For my next project I investigated the adaptive function of UV and fluorescent signals in H. erato (Chapter 3). I found that both signals are important for intraspecific signaling in Heliconius and are likely maintained through sexual selection, but under cloudy weather conditions, UV signals become ineffective whereas fluorescence persists as a reliable signal. Finally, I used behavioral tests to determine whether Heliconius have true color vision in the UV range as a consequence of UV opsin duplication (Chapter 4). I found that butterflies were capable of discriminating between more than one UV wavelength under varying light intensities, supporting that they have true UV color vision, which may aid in the detection of their genus-specific UV and fluorescent wing pigment. This dissertation (i) demonstrates how aposematic signaling may drive the evolution of social behavior in the context of visual ecology, (ii) shows how natural and sexual selection can work together to favor the evolution of specific animal phenotypes, and (iii) connects the evolution of specialized visual systems to specialized visual signals.