Skip to main content

Advertisement

SpringerLink
Log in

Can we perceive mental states?

  • Published:
Synthese Aims and scope Submit manuscript

Abstract

In this paper, I defend Non-Inferentialism about mental states, the view that we can perceive some mental states in a direct, non-inferential way. First, I discuss how the question of mental state perception is to be understood in light of recent debates in the philosophy of perception, and reconstruct Non-Inferentialism in a way that makes the question at hand—whether we can perceive mental states or not—scientifically tractable. Next, I motivate Non-Inferentialism by showing that under the assumption of the widely-accepted Principle of Cognitive Economy, any account that treats mental state perception as an inferential process commits itself to an unrealistically inefficient picture of our cognitive architecture. Drawing on research in cognitive science, I will then show that my Non-Inferentialist view receives direct support by the available empirical evidence. I conclude that there is no psychologically relevant sense in which perception of mental states differs from paradigmatic cases of perception, such as the perception of ordinary objects.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Source: Journal of Vision, which is open access

Similar content being viewed by others

Notes

  1. Conservative views about phenomenal content are defended, e.g., in Brogaard (2013), Byrne (2009), Prinz (2013), Sosa (2015) and Tye (1995, 2000). For liberal views on phenomenal content, see Bayne (2009), Block (2014), Butterfill (2009), Fodor (1983), Helton (2018), Nanay (2012), Newen (2017), Siegel (2010) and Vetter and Newen (2014). See also Helton (2016) for an overview.

  2. Several philosophers recognize that positions on the scope of “seeing” in the epistemic sense don’t entail commitments with regard to “seeing” in the technical sense of the perception-cognition debate (e.g., see Bayne 2009; Block 2014; Tye 1995).

  3. Some philosophers defend the view that seeing-as is not conceptual at all; these views can be found in Burge (2010), Orlandi (2011) and Gauker (2017). Space prevents me from addressing these views here; I’d just like to point out that insofar as it is assumed (as I do) that seeing-as is the perceptual competence to categorize or recognize stimuli, it is not clear to me how this is compatible with the claim that seeing-as is unconceptual. For a detailed discussion of the—in my view, insuperable—problems that nonconceptualist positions about seeing-as face, see Fodor (2015).

  4. Thus, if we understand “perception” in the sense employed in the debate about phenomenal content, the answer to the question I asked at the onset of this paper is possibly “no”.

  5. However, my view is still compatible with liberalism about phenomenal content. Thus, the liberal reader can understand my arguments as extending to the phenomenal sphere. In fact, I find myself convinced by the arguments for liberalism put forward in Bayne (2009), Fodor (1983), Helton (2018) and Siegel (2010).

  6. Roughly speaking, sensory transducers are devices that deliver us with raw sensory material.

  7. See Fodor (2005, p. 43): “It’s common ground for all versions of RTM that perception starts with an impression and ends with a categorization; that is, with the assignment of a concept. For example, it starts with an impression of this dog, and ends with the recognition of this dog; or of this dog as of a dog”.

  8. For examples of textbooks, see Marr (2010), Palmer (1999), Rolls and Deco (2001) or Snowden et al. (2012). For laboratories specializing on the semantic level of perception, see, e.g., the Stanford Computer Vision Lab, the Columbia CAVE lab, the Harvard Vision Sciences laboratory, or the Visual AI Lab at Princeton University.

  9. My reconstruction of the hypotheses made by Non-Inferentialists and the debate between Non-Inferentialists and Inferentialists differs from the one proposed in Spaulding (2015a). While her discussion is extremely insightful, it suggests that the two claims made by Non-Inferentialists—that we can perceive mental states and that mental state recognitions are non-inferential—are conceptually distinct and can’t be assimilated. By showing how the claims about inferential complexity and perception can be coherently bridged, I am offering a reconstruction that, in my opinion, recovers the theoretical intentions of participants in the debate in a way that is more accurate, fruitful and charitable.

  10. In fact, the principle plays an explanatorily central role in all domains of cognitive science, such as the study of language (Grice 1989; Leslie 2009), decision-making (Tversky and Kahneman 1983), attention (Posner 2012; Posner and Boies 1971), evolutionary psychology (Cosmides and Tooby 1994; Schulz 2011, 2016), and cognitive architecture more generally (Barlow 1961; Fodor 1983).

  11. For this reason, some cognitive scientists also prefer to call the concept we immediately apply in perceptual categorization ‘entry-level concept’.

  12. More generally, the facial expressions of the basic emotions reliably mark the emotions that cause them (Adolphs 2002; Ekman 1992; Ekman et al. 1987; Ohman and Soares 1993; see also Newen et al. 2015; Fasel and Luettin 2003; Tracy et al. 2015).

  13. Cf. also Newen et al. (2015) for arguing that basic emotions offer a stable, cross-cultural physiological pattern that can be reliably categorized as the emotion in question.

  14. For an argument for high-level representations of mental states in phenomenal content on the basis of expertise, see also (Newen 2017).

  15. Note that it is, for the reasons mentioned earlier, compatible with my position that System 1 or Core Cognition processes would possibly be classifiable as perceptual according to my criteria, while not be perceptual in the sense of the term employed in the perception-cognition debate. I’d like to thank an anonymous reviewer for raising this point.

  16. To accommodate this uncontroversial data, a common move made by conservatives about phenomenal content is to posit that high-level contents cause a difference in phenomenal content, but are not part of phenomenal content (see Bayne 2009; Block 2014; Tye 1995).

  17. Alternatively, it is also possible (and compatible with Non-Inferentialism) that there are two parallel events of perceptual categorization (e.g., of the face and the emotion).

  18. For a discussion of reverse inference in cognitive neuroscience, cf. Del Pinal and Nathan (2013) and Nathan and Del Pinal (2017).

  19. For an insightful discussion of the intricacies of the top-down/bottom-up distinction, see Shea (2015).

  20. Following the helpful characterization in Shea (2015), I here take “implicit representation” to refer to a “disposition to transition between two or more occurrent representations that can have no influence on subsequent processing except via the representations between which the disposition subsists” (p. 8), and “explicit representation” to refer to an occurrent (i.e. tokened) representation that is not implicit.

  21. See, e.g., Fodor (1983, 1985) and Pylyshyn (1999).

  22. Take “<anger>” to be an abbreviated description for the set of typical facial features exhibited by an angry person. In the rest of the paper, I will omit the angle brackets when referring to a type of facial configuration.

  23. An anonymous reviewer points out that in addition to the satisfaction of the four markers, a weak embodiment or physical realization basis of emotions might still be required in order to perceive something as a mental state. Otherwise, a scenario in which we see a light bulb that is attached to a safe and lighting up in different colors depending on which bank note values the safe’s sensor reads (e.g., green for 10 Euro, blue for 20 Euro, or red for 50 Euro), and apply monetary concepts in an automatic and fast fashion, would constitute an instance of perception, too. A plausible explanation of why it doesn’t constitute an instance of perception is that there is nothing that realizes properties sufficiently similar to Euro-properties as the basis of our visual input. Similarly, my argument would require that for direct perception of mental states, we need as sensory input enough features that are similar to those visual features typically realizing the mental state in question.

    I fully agree that given what we know about visual categorization, for a concept CF to be applied in perception, the stimulus must be sufficiently diagnostic for the presence of F, which includes similarity to other things that are F. Thus, in order to apply the concept anger, our cognitive system requires perceptual input that is sufficiently similar to typical tokens of anger. As a consequence, this requires a physical realization basis of anger in order to serve as appropriate visual input satisfying the similarity requirement. One account of such a basis can be given in terms of a pattern theory (see Newen et al. 2015).

    Although we might need a realization account like a weakly embodied pattern theory of emotions, it is not entirely clear whether we need an additional evidential measure, in addition to (i)–(iv), to capture this requirement. Note that the phenomenology-marker is arguably not satisfied in the thought experiment above. This can be explained by the fact that the realization-basis is lacking. In other words: if the realization-basis isn’t established, it might cause the lack of perceptual phenomenology; thus, evidential marker (i) might give us the realizer-requirement ‘for free’. Another question is whether there are ways to account for the posed problem on a purely epistemic (as opposed to metaphysical) basis. However, this is a question reserved for a future paper, as space prevents me from investigating this question in adequate detail here.

  24. Research suggests that this time-window can decrease after intense training (see Fabre et al. 2001).

  25. Here, research has especially been devoted to the investigation of the visual recognition of emotions. The focus on the mental state category of emotions rather than, say, intentions, is probably due to the more static nature of basic emotions as compared to intentional actions. Thus, cognitive measures of time-sensitivity can be applied more easily to emotional stimuli (faces with the respective expressions) than to action events that instantiate intentions, since those can hardly be presented in a static image.

  26. Note that by using the technique of reverse inference, I am not committing myself to a particular temporal boundary of perception. I simply use temporal information as an index for the presence of perceptual processes, in order to reverse-infer the presence of perceptual processes on the basis of the evidence.

  27. In a study testing contextual effects on emotion evaluation, Righart and de Gelder (2008) even found that contextual information can significantly reinforce N170 amplitudes. That is, a fearful face in a fearful scene elicited higher N170 effects than a fearful face shown in, say, happy scenes. Again, the theoretical lesson clearly is that early face processing stages have access to information of emotional categories. Otherwise, we would have a hard time explaining why a context assigned to the same emotional category reinforces N170 effects.

  28. I thank an anonymous reviewer for raising this point.

References

Download references

Author information

Authors and Affiliations

  1. University of Southern California, School of Philosophy, 3709 Trousdale Parkway, Los Angeles, CA, 90089-045, USA

    Eleonore Neufeld

Authors
  1. Eleonore Neufeld
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Eleonore Neufeld.

Additional information

Thanks to Guillermo Del Pinal, Alexander Dietz, Jeremy Goodman, Leif Hancox-Li, Daniel Pallies, Andrew Stewart, and audiences of USC’s Speculative Society, the iCog 4 in Oxford, and the “Perception and Justified Belief” conference in Bochum for helpful discussions and feedback on earlier versions of this paper. I owe special thanks to Mark Schroeder for extensive discussions and comments on multiple drafts. I’m also grateful to three anonymous reviewers for Synthese, who provided me with extremely valuable comments that led to key improvements of the paper.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Neufeld, E. Can we perceive mental states?. Synthese 197, 2245–2269 (2020). https://doi.org/10.1007/s11229-018-1807-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11229-018-1807-7

Keywords

Search

Navigation