Vector symbolic architectures are a viable alternative for Jackendoff s challenges

The authors, on the basis of brief arguments, have dismissed tensor networks as a viable response to Jackendoff’s challenges. However, there are reasons to believe that connectionist approaches descended from tensor networks are actually very well suited to answering Jackendoff’s challenges. I rebut their arguments for dismissing tensor networks and briefly compare the approaches. Van der Velde and de Kamps (V&K) have proposed neural blackboard architectures (NBAs) in response to Jackendoff’s (2002) challenges. Their note 1 dismisses tensor networks (Smolensky 1990) as a viable alternative. However, Gayler (2003) argues that vector symbolic architectures (VSAs)—connectionist approaches descended from tensor networks—are very well suited to answering Jackendoff’s challenges. There is not space here to repeat those arguments. Rather, I will rebut V&K’s arguments for dismissing tensor networks and briefly compare the approaches. Regardless of the ultimate relative success of NBAs and VSAs, the field of cognitive neuroscience will benefit from having plausible alternatives that can be compared and contrasted. V&K’s note 1 claims that “tensor networks fail to instantiate combinatorial structures.” Tensor product binding was developed specifically to address the issue of combinatorial id2704428 pdfMachine by Broadgun Software a great PDF writer! a great PDF creator! http://www.pdfmachine.com http://www.broadgun.com