Memorization of sequence-segments by humans and non-human animals: the Retention-Recognition Model Raquel G. Alhama, Remko Scha, Willem Zuidema Abstract: The existence of a statistical learning mechanism underlying the cognitive capacity to learn a “language” from an auditory input stream is well established. Computational models of segmentation formalize and test the specific theoretical assumptions about the mechanistic nature of this process. We present the Retention and Recognition model (R&R), a probabilistic model based on the cognitive processes of retention and recognition. We show that R&R outperforms other models in explaining for a range of experimental results: for a 2AFC task with human adults reported in Frank, Goldwater, Griffiths, and Tenenbaum (2010), as well as with data of human adults from a variant experiment from Peña, Bonatti, Nespor, and Mehler (2002), and with the responses of a segmentation experiment with rats (Toro & Trobalón, 2005). Our model also offers a new prediction on the response distribution over test items, which we will confirm revisiting these experimental results. Keywords: artificial language learning, language evolution, cognitive modelling