Ravens parallel great apes in flexible planning for tool-use and bartering

See allHide authors and affiliations

Science  14 Jul 2017:
Vol. 357, Issue 6347, pp. 202-204
DOI: 10.1126/science.aam8138

eLetters is an online forum for ongoing peer review. Submission of eLetters are open to all. eLetters are not edited, proofread, or indexed.  Please read our Terms of Service before submitting your own eLetter.

Compose eLetter

Plain text

  • Plain text
    No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g.
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Statement of Competing Interests

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Vertical Tabs

  • Flawed interpretations of planning behavior
    • Johan Lind, Associate professor, Centre for Cultural Evolution, Stockholm University

    In a recent study (Lind J 2018 What can associative learning do for planning? Royal Society Open Science, 5, 180778, [1]), I re-evaluated the study by Kabadayi and Osvath [2] taking the development of the ravens’ behavior into account. Here I address their conclusions, and conclude that behavior matching a current definition of flexible planning can emerge through our genetically guided associative learning model [3]. Therefore, ideas that ravens plan through mental simulations have little, if any, support.

    In the re-evaluation [1], computer simulations were performed based on procedural details of the raven study. The genetically guided learning model [3] was used to explore to what extent associative learning can explain the results found by Kabadayi and Osvath. Our learning model was able to account for the behavior of the raven (for details, see [1]). Correct choices in the test phases were found to depend on pre-training, in that the ravens were taught to choose the functional objects before testing started, and the fact that correct choices were rewarded throughout all test phases. The simulations show how both the value of choosing the functional objects and the reinforcing value of the functional objects themselves (conditioned reinforcement value) developed in pre-training and stayed high during all test phases.

    In addition, the simulations show that associative learning can explain why the ravens neglected small rewards in favor of the “neutral” obj...

    Show More
    Competing Interests: None declared.
  • RE: On rhetoric and empirical science
    • Mathias Osvath, Associate professor in Cognitive Zoology, Lund University Cognitive Science
    • Other Contributors:
      • Can Kabadayi, Postdoctoral Researcher, Lund University Cognitive Science

    In their eLetter Redshaw et al reiterate their critique [1] which we have responded to in some detail [2]. At the same time, they avoid replying to our concerns about their inadequate reasoning.

    Redshaw et al contend that a fundamental problem with our study is a statement in the fact box in our response [2] (not the original paper): learning to use a tool by observation, and displaying the skill five times without errors, is observational learning. We maintain this standard definition, which had no bearing on the conclusions of our study – those were based on experimental controls. Hardly a fundamental problem.

    Ignoring these controls, Redshaw et al say we should have shown that the ravens’ preferences for objects are based on the future they had reason to predict. We did this, and incorporated a sine qua non of planning: a temporal perspective. Our control gave very clear results [2,3]. Redshaw et al still argue that crucial behaviors could be explained with hitherto unknown forms of associative learning (or – incomprehensibly – with a planning mechanism).

    Redshaw et al regard our work as a matter of faith, semantics and rhetoric, and oddly enough call for empirical research. The study is one of the most extensive studies of animal planning to date, based on decades of other research and theorizing. Redshaw et al forward alternative explanations with functions not described in the literature, and refrain from explaining how they even hypotheticall...

    Show More
    Competing Interests: None declared.
  • Clarifying animal planning capacities
    • Jonathan Redshaw, Postdoctoral Research Fellow, School of Psychology, University of Queensland
    • Other Contributors:
      • Alex H Taylor, Senior Lecturer, School of Psychology, University of Auckland
      • Thomas Suddendorf, Professor, School of Psychology, University of Queensland

    We would like to direct readers to an ongoing debate about Kabadayi and Osvath’s study [1]. We recently pointed out why the results are not entirely compelling, and suggested experimental avenues towards resolving the issues [2]. In their reply [3], the authors sidestep the major concerns we raise and instead portray our critique as one driven by preconceived ideas and disagreements about semantics. So we would like to clarify here the most fundamental problem with their study:

    The ravens first learned that selecting a target object can bring about food rewards. The authors claim that the initial learning was observational and not associative [3], yet the birds had at least 5 training trials of associating the object with rewards. An associative learning perspective would therefore strongly predict ravens to have a significant preference for the target object over objects that have not been associated with reward [2].

    The authors [3] do not consider our constructive suggestions [2] that could clarify the competences of ravens, such as whether they can change their preferences for objects based on which particular future problem they have reason to predict will reoccur. Whether, and if so, to what extent, ravens can flexibly plan is not an article of faith, but a matter of testing the null hypothesis, systematically ruling out alternative explanations, and independent replication [4,5].

    Contrary to what the authors imply, we currently do not have a...

    Show More
    Competing Interests: None declared.

Stay Connected to Science

Navigate This Article