The thesis of this paper emerges from the fact that mental states are generated by neural processes that also produce an associated electroencephalogram (EEG). Thus, it is logical to expect correlations between mental state and EEG. The corollary is that the EEG can serve as an index of mental state, which can be particularly useful for studies in animals, where mental states are much less accessible for objective study than in humans. Herein, I briefly review the traditional approaches that have informed our attitudes about animal mental states. Virtually all of our conclusions about mental states in animals are drawn by inference from behavioral observation, a process that is highly and unavoidably subject to anthropomorphism. Traditionally, the electroencephalogram (EEG) has been used in a crude way as an objective indication of physical and behavioral state in animals. This, however, has led to substantial controversy, because there are several situations in which EEG patterns and behavior seem to be dissociated. We not only fail to understand these dissociated states, but there are also important humane animal-welfare issues that remain unresolved because we do not fully understand the extent to which the EEG can reflect mental state. At issue is whether EEG-behavioral dissociations, to the extent that they exist, are proof that the EEG is dissociated from mental states. Powerful new EEG methods, such as topographical EEG mapping, wavelet analysis, and testing for nonlinear (‘chaotic’) dynamical properties and short-term serial dependencies, are now available for studying the extent to which the EEG can index thinking and feeling in humans and, by extrapolation, in animals. Critics who have become disenchanted with the utility of the EEG should at least concede that fresh approaches to old problems are now available and should therefore be thoughtfully considered. If such research does nothing more than improve the rigor of the debate over animal welfare and rights issues, it will be worth the effort.