Abstract
Cognitive scientists use computational models to represent the results of their experimental work and to guide further research. Neither of these claims is particularly controversial, but the philosophical and evidentiary statuses of these models are hotly debated. To clarify the issues, I return to Newell and Simon’s 1972 exposition on the computational approach; they herald its ability to describe mental operations despite that the neuroscience of the time could not. Using work on visual imagery (cf. imagination) as a guide, I examine the extent to which this holds true today. Does contemporary neuroscience contain mechanisms capable of describing experimental results in imagery? I argue that it does not, first by exploring foundational achievements in imagery research then by showing that their neural basis cannot be specified. Newell and Simon’s methodological position accordingly stands, even 50 years later. Computational — as opposed to physiological — descriptions must be retained to characterize and study mental phenomena, even as we learn high-level details of their implementation via brain data.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
This is not to deny that some authors avowed “in-principle” autonomy between the disciplines. They certainly did, though Human Problem Solving appears to have the correct idea.
I will adopt Newell and Simon’s loose usage of the term information processing for the purposes of this paper. I’m interested in exploring the “symbolic model[s] on the basis of which pertinent specific aspects of… behavior can be calculated.” Whether these systems are truly functionalist or information processing systems in the technical senses of these terms is not germane to the questions I’m considering. See Piccinini’s article (Piccinini and Scarantino 2010) for discussion.
Imagery might first be likened to imagination, but seeing as the study of imagery has come to encapsulate involuntary activation of the visual system (as in, for instance, pattern recognition), we cannot rely on the colloquial term too heavily.
In addition to improved behavioral measures and prevailing perceptual theory.
This is not to suggest that impermanent knowledge exists. A claim shown to be false was never known in the first place (Azzouni 2020).
“Image” here simply means “form a mental image.”
The experiment is described in greater detail in Kosslyn et al. (1983).
This is what makes the present experiment so solid: it is impervious to possible effects of factors like image density.
Importantly, the constraints imposed on explanation by any single experiment are far less severe than those of an entire program. The scope of Kosslyn’s work is considerably broader than the isolated results I review in detail here—the image generation literature alone has many robust findings (e.g., Kosslyn 1994, p. 294-5).
The experiment is described in greater detail in Kosslyn et al. (1977).
This took me a moment to wrap my head around. An example helped. If this model uses gross category information to distinguish the very largest and very smallest stick figures (so the largest in the large group, the smallest in the small group), this should take the same amount of time as distinguishing the second largest and second smallest, as in each case all that must be compared is gross category information. The same is true of comparisons within the detailed group; no size-disparity effects are expected for comparing the smallest and second smallest, as opposed to the smallest and third smallest figures. Needless to say, this is not so for the imagistic models.
There is a latent assumption here that the relevant categories cognitively are the learned ones. The results we will observe render this assumption unproblematic.
1-5 in the previous section are similarly coarse inferences from underlying mechanism to overt behavior. More precision, closer to the kind familiar from developed sciences (Smith 2014), can be achieved by tuning cognitive theory to measured parameters within individuals, the approach of Kosslyn et al. (1984). This is surely more productive than the approaches to individual differences common in the psychology literature, namely developing survey-based measurements of “vividness,” “control,” and other adjectives whose cognitive correlate isn’t clear.
Somewhat more subtly, to study image “generation” is to presume some distinction between stored representations and those being actively entertained.
I don’t make this point to berate developmental linguists in the generative tradition any more than they already have been (e.g., Christiansen and Chater 2008, 2009). Instead my point is to suggest that carrying on research in pursuit of UG’s properties, as opposed to merely its existence, may be both evidentially productive and more effective at meeting criticism than the current strategy.
Similarly, “Most research on predictive language processing in the last 15-20 years has focused on demonstrating that prediction is an important part of language processing. Much less research has been directed at establishing the mechanisms and mediating factors of anticipatory language processing” (Huettig 2015).
Kosslyn’s cumulative approach described here is not unique in the history of psychology. Consider, for instance, Hubel and Wiesel’s early exploration of the critical period in cat and monkey visual development. They gained access to the properties of this period “by closing one eye at different ages and keeping it closed for several months or longer” then measuring, for example “the relative influence of the two eyes on single cortical cells” (Wiesel 1982). Later, via autoradiography measurements taken from a monkey whose monocular deprivation was switched from one eye to another (at 3 weeks), they discovered “the critical period is different for the two cell types. Whereas the critical period is over for the magnocellular input at 3 weeks, the parvocellular input apparently begins to lose its ability to expand at 6 weeks” (Wiesel 1982). Their ability to more finely resolve these details of the critical period (and others, e.g. “competitive mechanisms rather than disuse are prime factors in producing the changes observed under conditions of monocular deprivation.”) is similarly enabled by large numbers of presuppositions (which are indirectly tested) about the visual system and its development.
I will eventually point out that much contemporary research fails to engage with cognitive theory, to its detriment.
It’s a little misleading to refer to to theory development and testing as though they are independent. As I have stressed in this paper, theory components are subject to stringent test by subsequent theory development.
I thank George Smith for this point.
This may be somewhat understated. Changes to the transformational component of the theory may force alterations to other components.
The (distinct) question of why researchers (perhaps journals) feel obliged to title their papers as such, neglecting the more accurate “possible localization of cognitive function x” is a sociological one. To speculate, the prefix “-neuro” and techniques from cognitive neuroscience may function honorifically in the literature. See Weisberg et al. ’s (2008) “The Seductive Allure of Neuroscience Explanations” for some related discussion.
Chomsky cites Gallistel to this effect; see, for instance, Berwick and Chomsky (2016, p. 50-2).
Recall the “common point in the philosophy of science” to which Smith and Kosslyn referred in 1980: “empirical research in a theoretical vacuum is likely to flounder” (Smith and Kosslyn 1980).
It may be that researchers emphasize as they do because they suspect neuroimaging techniques will eventually unveil in great detail the workings of the mind, but this is a paper about evidence-based achievements in the present.
See Buchwald and Smith (2001) for a discussion of the difference between conceptual and evidentiary revolutions.
References
Azzouni, Jody. 2020. Attributing Knowledge: What it Means to Know Something. Oxford University Press.
Bédécarrats, Alexis, et al. 2018. RNA from trained Aplysia can induce an epigenetic engram for long-term sensitization in untrained Aplysia. Eneuro 5(3).
Beniaguev, David, Idan Segev, and Michael London. 2021. Single cortical neurons as deep artificial neural networks. Neuron 109 (17): 2727–2739.
Berwick, Robert C., and Noam Chomsky. 2016. Why only us: Language and evolution. MIT press.
Buchwald, Jed Z., and George E. Smith. 2001. Incommensurability and the discontinuity of evidence. Perspectives on Science 9 (4): 463–498.
Chang, Le., and Doris Y. Tsao. 2017. The code for facial identity in the primate brain. Cell 169 (6): 1013–1028.
Chomsky, Noam. 2000. New Horizons in the Study of Language and Mind. Cambridge University Press.
Christiansen, Morten H., and Nick Chater. 2008. Language as shaped by the brain. Behavioral and Brain Sciences 31 (5): 489–509.
Christiansen, Morten H., and Nick Chater. 2009. The myth of language universals and the myth of universal grammar. Behavioral and Brain Sciences 32 (5): 452–453.
Coltheart, Max. 2006. What has functional neuroimaging told us about the mind (so far)? In European Cognitive Neuropsychology Workshop, 2005, Bressanone, Italy; Position paper presented to the aforementioned conference. Masson Italia.
Damasio, Antonio R., and Norman Geschwind. 1984. The neural basis of language. Annual Review Of Neuroscience 7 (1): 127–147.
Decety, Jean. 1996. The neurophysiological basis of motor imagery. Behavioural Brain Research 77 (1–2): 45–52.
Denny, Christine A., Evan Lebois, and Steve Ramirez. 2017. From engrams to pathologies of the brain. Frontiers in Neural Circuits 11: 23.
Dijkstra, Nadine, Sander E. Bosch, and Marcel A.J.. van Gerven. 2019. Shared neural mechanisms of visual perception and imagery. Trends in Cognitive Sciences 23 (5): 423–434.
Gallistel, Charles R. 2017. The coding question. Trends in Cognitive Sciences 21 (7): 498–508.
Gallistel, Charles R. 2021. The physical basis of memory. Cognition 213: 104533.
Gold, Joshua I., and Michael N. Shadlen. 2007. The neural basis of decision making. Annual Review of Neuroscience 30: 535–574.
Harrington, Deborah L., and Kathleen Y. Haaland. 1999. Neural underpinnings of temporal processing: A review of focal lesion, pharmacological, and functional imaging research. Reviews in the Neurosciences 10 (2): 91–116.
Hubel, David H. 1982. Exploration of the primary visual cortex, 1955–78. Nature 299 (5883): 515–524.
Huettig, Falk. 2015. Four central questions about prediction in language processing. Brain Research 1626: 118–135.
Jackendoff, Ray S. 2002. Foundations of language: Brain, meaning, grammar, evolution. USA: Oxford University Press.
Kosslyn, Stephen M., et al. 1977. Category and continuum in mental comparisons. Journal of Experimental Psychology: General 106 (4): 341–375.
Kosslyn, Stephen M. 1980. Image and Mind. Harvard University Press.
Kosslyn, Stephen M., et al. 1983. Generating visual images: Units and relations. Journal of Experimental Psychology: General 112 (2): 278–303.
Kosslyn, Stephen M., et al. 1984. Individual differences in mental imagery ability: A computational analysis. Cognition 18 (1–3): 195–243.
Kosslyn, Stephen M. 1994. Image and brain: The resolution of the imagery debate. MIT press.
Kosslyn, Stephen M., et al. 2001. Imagining rotation by endogenous versus exogenous forces: Distinct neural mechanisms. NeuroReport 12 (11): 2519–2525.
Kosslyn, Stephen M., Giorgio Ganis, and William L. Thompson. 2001. Neural foundations of imagery. Nature Reviews Neuroscience 2: 635–642.
Martin, Alex, and Jill Weisberg. 2003. Neural foundations for understanding social and mechanical concepts. Cognitive Neuropsychology 20 (3–6): 575–587.
Mayseless, Naama, Ayelet Eran, and Simone G. Shamay-Tsoory. 2015. Generating original ideas: The neural underpinning of originality. Neuroimage 116: 232–239.
Miller, Brett, Neil Myler, and Bert Vaux. 2016. Phonology in universal grammar. In The Oxford Handbook of Universal Grammar.
Moeller, Sebastian, Winrich A. Freiwald, and Doris Y. Tsao. 2008. Patches with links: A unified system for processing faces in the macaque temporal lobe. Science 320 (5881): 1355–1359.
Newell, Allen, and Herbert A. Simon. 1972/2019. Human Problem Solving. Echo Point Books and Media. isbn: 978-1-63561-792-4.
Paivio, Allan. 1975. Perceptual comparisons through the mind’s eye. Memory & Cognition 3 (6): 635–647.
Perky, Cheves West. 1910. An experimental study of imagination. The American Journal of Psychology 21 (3): 422–452.
Piccinini, Gualtiero, and Andrea Scarantino. 2010. Computation vs. information processing: Why their difference matters to cognitive science. Studies in History and Philosophy of Science Part A 41 (3): 237–246.
Piccinini, Gualtiero. 2020. Neurocognitive Mechanisms: Explaining Biological Cognition. Oxford University Press.
Poeppel, David, and M. Florencia Assaneo. 2020. Speech rhythms and their neural foundations. Nature Reviews Neuroscience 21 (6): 322–334.
Pylyshyn, Zenon W. 1984. Computation and Cognition. MIT press.
Rilling, James K., et al. 2002. A neural basis for social cooperation. Neuron 35 (2): 395–405.
Shrout, Patrick E., and Joseph L. Rodgers. 2018. Psychology, science, and knowledge construction: Broadening perspectives from the replication crisis. Annual Review of Psychology 69: 487–510.
Smith, George E., and Stephen M. Kosslyn. 1980. An information-processing theory of mental imagery: A case study in the new mentalistic psychology. In PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association. Vol. 1980. 2. Philosophy of Science Association, 247–266.
Smith, George E. 2010. Revisiting accepted science: The indispensability of the history of science. The Monist 93 (4): 545–579.
Smith, George E. 2014. Closing the loop. In Newton and Empiricism, 262–352.
Smith, George E., and Raghav Seth. 2020. Brownian Motion and Molecular Reality: A Study in Theory-Mediated Measurement. Oxford University Press.
Spitzer, Hedva, Robert Desimone, and Jeffrey Moran. 1988. Increased attention enhances both behavioral and neuronal performance. Science 240 (4850): 338–340.
Turing, Alan Mathison. 1936. On computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society 2: 230–265.
Wagenmakers, Eric-Jan. 2007. A practical solution to the pervasive problems of p values. Psychonomic Bulletin & Review 14 (5): 779–804.
Weisberg, Deena Skolnick, et al. 2008. The seductive allure of neuroscience explanations. Journal of Cognitive Neuroscience 20: 470–477.
Wiesel, Torsten N. 1982. Postnatal development of the visual cortex and the influence of environment. Nature 299 (5884): 583–591.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Caulfield, A. Symbol and Substrate: A Methodological Approach to Computation in Cognitive Science. Rev.Phil.Psych. (2024). https://doi.org/10.1007/s13164-023-00719-4
Accepted:
Published:
DOI: https://doi.org/10.1007/s13164-023-00719-4