2500 words

What is the multiple realizability argument?

The multiple realizability argument (MRA) is an argument directed at type-identity theories of mind, while being used for and against functionalist theories of mind. (I think Ross’ 1992 argument in Immaterial Aspects of Thought and Feser’s 2013 arguments refute functionalist theories.) First formulated by Hilary Putnam in 1975, the argument he formulated can be put like this:

P1. If type-physicalism is true, then every mental property can be realized in exactly one physical way.
P2. It is empirically highly plausible that mental properties are capable of multiple realizations.
C1. It is (empirically) highly plausible that the view of type-physicalism is false (modus tollens, P1, P2). (From Just the Arguments, 81. Putnam’s Multiple Realization Argument against Type-Physicalism)

P1 states the scope of type-physicalism—that all mental states are realizable in one and only one physical way. P2 states that it is probable that mental properties are capable of multiple realizations. This premise is an empirical one, and so we need evidence to believe it. Then, the conclusion is that type-physicalism is false sine mental states are multiply realizable. Quite obviously, this argument shows that mental states don’t reduce to brain states, which means that physicalism is false. Since P2 needs defense, I will defend it in this article while giving my own formulation of the MRA.

Here is my formulation of it:

P1: If mental properties were identical to physical properties, then any change in one entails a change in the other.
P2: Mental properties can change without any corresponding change in physical properties.
C: So mental properties aren’t identical to physical properties.

As you can see, like Putnam’s formulation, P2 is an empirical claim and so needs empirical support. So if one mental state can be realized in multiple ways, then type-physicalism (mind-brain identity) is false. I will spend the rest of this article arguing for the truth of P2 and then provide an argument from analogy showing that mental states are multiply realizable.

An a priori argument for multiple realizability

If the MRA were true, then there would be evidence of a specific mental state that is realized in multiple physical ways. While empirical evidence is irrelevant to metaphysical possibility (and to concepts), multiple realizability can be known a priori. Before I give the empirical argument from analogy for multiple realizability, I will give the a priori argument.

P1: Mental states and processes exhibit certain characteristic features and properties like intentionality, subjectivity, and causality.
P2: If mental states are multiply realizable, then they are not reducible to their underlying physical properties.
C: Thus, mental states and processes are not reducible to their underlying physical properties.

Or

P1: M has properties P1, P2, P3…
P2: If M is multiply realizable, then M is not reducible to it’s physical properties.
C: Therefore, M is irreducible to its physical properties.

Premise 1: Mental states and processes are characterized by what they do rather than what they’re “made of.” Intentionality is the ability for mental states to be “about” things, while directed at objects, events or states of affairs like when a belief or proposition is about a certain end goal. So M properties aren’t reducible to any P properties, and intentionality is a property of mental states which set them apart from physical states, since purely physical things can never have the ability to intend. Subjectivity refers to the fact that mental states are experienced through a first-personal perspective which can’t be observed or measured by others. This property sets M states apart from P states, since physical states can be studied and observed from a third-personal perspective. So while we can study brain states, since mental states don’t reduce to brain physiology, then by studying brain states we aren’t studying the mind. Lastly the property of causality refers to the fact that mental states and processes have causal effects on action and behavior, cognition and other mental states and processes. So the distinctive role that mental states and processes play in generating action, behavior, and cognition cannot be captured by studying the brain or the body.

Premise 2: This premise highlights the fact that multiple realizability implies that mental states can be realized in a multitude of physical states and processes without any loss of mental properties. So the Conclusion follows that mental states and processes are irreducible to their underlying physical properties.

So if mental states and processes have characteristic features that distinguish them from other kinds of states and processes, and if they can be realized by multiple physical systems, then they cannot be reduced to one physical system.

Defending P2: Empirical evidence for multiple realizability

The way that Putnam and I have formulated the argument for MR is an empirical claim. So empirical claims require empirical evidence. While the previous argument was a priori, it could therefore be argued without empirical evidence.

The example of visual perception. Most animals on earth have vision. The visual systems of animals have evolved to help them survive in their ecologies. Humans have three cone types in their eyes which allows them to see a range of colors. On the other hand, some birds have four cone types which allow them to see a wider range of color than humans, and the fourth cone thsg birds have allows them to see more colors than humans (Stoddard et al, 2020). Bats have evolved eyesight that allows them to see in low light environments, while eagles have evolved eyesight that allows them to see objects at great distances. So despite differences in the visual systems between animals, they can all recognize objects and visually navigate their ecologies. So different animals have evolved different vision systems that help in a certain niche. Furthermore, different types of photoreceptors have evolved in different animals, with different connections between the eye and the brain, which began evolving around 600 million years ago, with the Cambrian explosion leading to body plans and systems which then supported vision (Lamb, Pugh Jr, and Collin, 2008; Lamb, Collin, and Pugh Jr, 2011; Asteriti, 2015). These photoreceptors come in two kinds—ciliary type (c-type) or rhabdomeric type (r-type); vertebrates seek to have a unique mix of these cone types which allow a wide range of vision (Marshedian and Fain, 2017). Certain eye structures have also evolved independently (Land and Nilsson, 2012). So different animals have different numbers of photoreceptors and cones, which help them to visualize their environment; the diversity of rods and photoreceptors in the animal kingdom is vast (Piechl, 2005). Thus, the evidence cited here shows that different animals have differently-evolved visual systems, but they can still visualize their environments even though the physical systems that allow it are different.

The brain’s ability to compensate from injury and the brain’s of athletes and musicians. After a traumatic brain injury occurs, the brain—being plastic—can rewire itself to carry different functions after an injury. For example, in the blind, the visual cortex is repurposed and processes tactile and auditory stimuli (Elbert et al, 2002; Lane et al, 2015; Gori et al, 2019). The primary motor cortex in musicians is larger than non-musicians, and this is due to constant practice on their instrument of choice (Toyka and Freund, 2006; Watson, 2006; Olszewska et al, 2021). Basketball players have larger cortical areas associated with visual processing and attention (Kim et al, 2022) along with athletes having different cortical neuronal networks than novices (Tan et al, 2017). This then is solid evidence for the claim that learning new skills and continously performing them at an expert level leads to changes in the brain (Park et al, 2015). Further, when it comes to the brain’s ability to heal from an injury, it has been shown that if a certain brain area is impacted, other parts of the brain will pick up the slack of the injured part, which shows the plasticity of the brain and the brain’s ability to compensate for an injury to it by directing and making new neural pathways to carry out new tasks (Nishimura et al, 2009; Su, Veeravuga, and Grant, 2016; Hylin, Kerr, and Holden, 2017). Thus, the evidence cited here shows how the brain can adapt to tasks that a person performs, and how it can adapt to changes to it (like injury) and even repurposing certain parts of itself in people with certain disabilities. This, like the example of visual perception, lends further support for the claim that different physical systems can perform the same mental function.

Brain-computer interfaces. Lastly, we have brain-computer interfaces. These interfaces “acquire brain signals, analyze them, and translate them into commands that are relayed to output devices that carry out desired actions” (Shih et al, 2012). These interfaces allow humans to control things with their thoughts, bypassing the need for physical interfaces; this technology also allows individuals to control certain kinds of devices using brain waves using their mental intentions (Mak and Wolpaw, 2010). People with these interfaces can control prosthetic limbs (Mischenko et al, 2017; Murphy et al, 2017; Asanza et al, 2022). These interfaces have also been explored to give people the ability to communicate with speech again (Brumberg et al, 2010). This also supports MR since brain-computer interfaces which use EEG to record brain activity could translate mental states into movements while interfaces that use implanted electrodes may allow an individual to control a robotic arm. Thus, the development of this technology shows that different mental states can be realized in different physical systems which is then dependent on the type of interface used.

Strappini et al (2020) provide yet more empirical support for MR. They write:

Here, we illustrate some cases that provide empirical evidence in support of MRT. Recently, it has been proposed that foveal agnosic vision, like peripheral vision, can be restored by increasing object parts’ spacing (Crutch and Warrington, 2007; Strappini et al., 2017b). Agnosic fovea and normal periphery are both limited by crowding, which impairs object recognition, and provides the signature of visual integration. Here, we define a psychological property of restored object identification, and we cross-reference the data of visually impaired patients with different etiologies. In particular, we compare the data of two stroke patients, two patients with posterior cortical atrophy, six cases of strabismic amblyopia, and one case with restored sight. We also compare these patients with unimpaired subjects tested in the periphery. We show that integration (i.e., restored recognition) seems to describe quite accurately the visual performance in all these cases. Whereas the patients have different etiologies and different neural correlates, the unimpaired subjects have no neural damage. Thus, similarity in the psychological property given the differences in the neural substrate can be interpreted in relation to MRT and provide evidence in its support

While Booth (2018: 143-144) uses the example of multilingualism to support MR:

First, there are multiple ways of speaking a second-language, based on difference between high proficiency early and late bilinguals. Second, there are multiple ways of being a speaker of a given language, specifically as a monolingual or bilingual speaker of that language, where the language is the bilingual speaker’s L1. These examples meet the conditions advanced by Polger and Shapiro for examples of multiple realization, and should therefore be accepted as genuine cases of multiple realization.

Now that I have given a good overview of the evidence in support of MR, I will not provide the argument.

The empirical argument from analogy for multiple realizability

P1: Different animals have evolved different vision systems to suite their ecologies.
P2: Humans have a trichromatic visual system while some birds have tetrachromatic visual system while some insects have compound eyes.
P3: Despite differences in these visual systems, these animals all are able to perform similar visual tasks, like spatial navigation and object recognition.
P4: Studies of brain damage and neuroplasticity show that different brain regions can take on different functions after injury or training, like blind people using the visual cortex for auditory processing, muscisians having larger motor areas for finger control, and basketball players having larger cortical areas associated with visual processing and attention.
P5: The development of brain-computer interfaces show that mental states can be translated into different forms of output, like movement, speech, and text, using different physical devices.
C: Thus, multiple realizability is true, since the mental state of visual perception (and other mental states) can be realized in different physical systems without affecting functioning.

---

P1: If mental states can only be realized in a single physical system, then all animals with similar cognitive tasks should have identical neural structures.
P2: Different animals have different neural structures for performing similar cognitive tasks, like visual perception.
C: Thus, mental states cannot be realized in a single physical system.

---

P1: If mental states can only be realized in a single physical system, then all animals with similar cognitive tasks should have identical neural structures.
P2: If all animal with similar cognitive tasks have identical neural structures, then different animals should not have different neural structures for performing similar cognitive tasks.
P3: Different animals have different neural structures for performing similar cognitive tasks, like spatial navigation, object recognition and visual perception.
C: Thus, mental states cannot be realized by a single physical system.

Conclusion

I have provided both an a priori and a posteriori argument for the MRA. The a priori argument shows that multiple realizability is metaphysically possible, while the empirical evidence I have cited along with the empirical premises in my arguments have shown that multiple realizability is an empirically defensible position. It seems to me that it is intuitive that different mental states can be realized by different physical systems and not only one kind of physical system.

The a priori argument shows that mental states and physical states have different properties; physical states cannot have the properties that mental states have. So this shows that it’s metaphysically possible the MR is true, while the empirical evidence and arguments I have mounted show that it is true in our world as well. Mental states can’t be reduced to physical states, mental states are causally efficacious, and there are multiple ways to achieve the same cognitive function, like visual perception across the animal kingdom. The example of visual perception of different animals, studies of athletes, musicians, and people with traumatic brain injuries, and even brain-computer interfaces show that different mental states can be realized in multiple physical ways.

So if this is true, then multiple realizability is true. If multiple realizability is true, then type-physicalism is false, and therefore identity theories of mind need to find another avenue to prove their thesis. Mind-brain identity is clearly false; mind doesn’t reduce to brain and mental states can be realized by different physical systems. This is yet another argument against physicalism—the attempted reduction of mind to brain. Physicalism is quite clearly false.