Traditionally, studies of delayed matching-to-sample (DMTS) tasks in nonhuman species have focused on the assessment of the limits of the retrieval of information stored in short-and longterm memory systems. However, it is still unclear if visual recognition in these tasks is affected by very brief delay intervals, which are typically used to study rapidly decaying types of visual memory. This study aimed at evaluating if tufted capuchin monkeys' ability to recognise visual stimuli in a DMTS task is affected by (i) the disappearance of the sample stimulus and (ii) the introduction of delay intervals (0.5, 1.0, 2.0 and 3.0 seconds) between the disappearance of the sample and the presentation of the comparison stimuli. The results demonstrated that the simple disappearance of the sample and the introduction of a delay of 0.5 seconds did not affect capuchins' performance either in terms of accuracy or response time. A delay interval of 1.0 second produced a significant increase in response time but still did not affect recognition accuracy. By contrast, delays of 2.0 and 3.0 seconds determined a significant increase in response time and a reduction in recognition accuracy. These findings indicate the existence in capuchin monkeys of processes enabling a very accurate retention of stimulus features within time frames comparable to those reported for humans' sensory memory (0.5-1.0 seconds). The extent to which such processes can be considered analogous to the sensory memory processes observed in human visual cognition is discussed. We examine capuchin monkeys' ability to match visual stimuli after brief delays. 0.5 sec intervals do not affect either response accuracy or response time.  1.0 sec intervals do not affect response accuracy but increase response time. 2.0 and 3.0 sec intervals decrease response accuracy and increase response time. Effects consistent with those ascribed to iconic memory in humans. Research Highlights Responses to the ReviewerWe were extremely pleased that the reviewer was satisfied with the vast majority of our revisions and that s/he felt that we had been very responsive and addressed most of her/his concerns. We are also pleased to read that the reviewer felt that all her/his specific comments have been addressed.We have now further revised the article in several places (P 4 lines 46-48; P 6 lines 53-59; P 7 lines 1-3; P 13 lines 44-49; P 14 lines 1-7; P 18 lines 48-54; P 18 line 56 -P 19 line 1) in order to make it absolutely clear that we make no claims regarding the high-capacity of the memory system that we have observed in capuchin monkeys so that, in that respect it could be considered analogous to human iconic memory. All our claims are based on analyses of the effects of time delay which, because they suggest a discontinuous function of the delays, can be considered indicative of separate memory processes at early stages of the encoding of visual information in monkeys.As suggested by the reviewer we now present effect sizes for paired t tests and 95% confidence in...

Acknowledgements 21We thank two anonymous reviewers for their thoughtful suggestions and constructive comments on 22 the manuscript. We are also grateful to Tyrone Lucon Xiccato for his helpful statistical suggestions.

All authors contributed equally to the paper.We wish to thank the three anonymous reviewers for their valuable suggestions and constructive comments on the manuscript. We acknowledge Massimiliano Bianchi and Simone Cartarinacci for help with animal management and Ludovica Cazzato for help with data collection. We also thank the Comune di Roma-Museo Civico Zoologia and Non-human primates, differently from humans, are less proficient at processing 6 global properties of visual compound stimuli. It has been suggested that humans 7 preferentially process stimuli globally because this enables a more economical encoding 8 of the stimuli. In this study we assessed the role of short-term memory in global/local 9 processing by presenting tufted capuchin monkeys with Navon-type hierarchical figures 10 in both simultaneous and delayed matching-to-sample tasks. Capuchins' ability to 11 discriminate hierarchical stimuli was evaluated as a function of increasing delay intervals 12 (0.0, 0.5, 1.0, 2.0 and 3.0 sec) between the disappearance of the sample and the 13 presentation of the comparison stimuli. The results showed that recognition accuracy for 14 local features was above chance level with delays of up to 3.0 sec, as previously reported 15 when capuchins were faced with non-hierarchical stimuli. By contrast, the recognition of 16 global configurations was above chance level in simultaneous, 0.0 and 0.5 sec delay 17 conditions but not at delay intervals of 1.0 sec or longer. These findings indicate that 18 capuchins' propensity to process the local properties of visual stimuli can be observed 19 when a delay is interposed between the presentation of sample and comparison stimuli 20 and was not reversed by increasing the delay. Moreover, our results show that capuchins' 21 local propensity was not reversed by increasing stimulus size. Overall, our study 22 confirms crucial differences between human and non-human primates and adds new 23 insights into the comparative research on visual grouping functions of these species. 24 25