Neutral Monism and the Attributes of God in Spinoza’s Metaphysics

1 Introduction

Spinoza’s conception of attributes has been a central topic of dispute among his readers since the seventeenth century.^[1] In Spinoza’s system, as in Descartes’, the attributes of thought and extension map onto the distinction between the mind and the physical world. Just like Descartes’ “principal attributes” (Descartes 1985, 1:53), Spinoza’s attributes are the most general properties of entities, that is, those that define their essence, whereas their more specific properties are ‘affections’ or ‘modes’ of these attributes (EP25C).^[2] Yet Spinoza departs in important ways from the Cartesian notion of a principal attribute. For Spinoza, the attributes of thought and extension are not mutually exclusive: he argues there is only one substance (E1P14) which is a thinking thing and an extended thing (E2P1–2), and this holds true of each of its modes as well (E2P7S). Moreover, Spinoza claims there must be infinitely many attributes, corresponding to the infinite ways in which God’s intellect can conceive of the nature of the one substance (i. e., God’s own essence; E1D6, E1P16).

Descartes (1985, 1:53) defines the principal attribute of a substance as that which constitutes its essence. Spinoza’s definition nearly mirrors Descartes’, but for one notable amendment: he defines an attribute as “what the intellect perceives of a substance, as constituting its essence” (E1D4). This is the only definition in Spinoza’s Ethics that is phrased in terms of what the intellect perceives, leading a number of interpreters to develop an attribute-neutral reading of his metaphysics based on this definition. On this reading, substance is, in itself, undifferentiated with respect to attributes; these differences come into play only insofar as the substance is perceived by the intellect. Given Spinoza’s commitment to substance monism (E1P14), we can describe the resulting position as neutral monism, the view that the world is constituted by a single substance which is neither thinking nor extended.^[3] The most extreme version of the attribute-neutral reading is defended by Henry Wolfson (1958, 1:5), who situates Spinoza in the tradition of negative theology, especially as regards Jewish and Arabic mediaeval thought.^[4] He refers to his reading as a ‘subjectivist’ account of the attributes, according to which they are “invented by the mind” rather than “discovered by the mind” (Wolfson 1958, 146).

The subjectivist reading of Spinoza has been met with sharp criticism since Wolfson’s seminal work.^[5] Many scholars have instead opted for a thoroughly objectivist reading, according to which God’s essence is constituted by a plurality of distinct attributes.^[6] Others have attempted to escape the dichotomy between objectivism and subjectivism by adopting intermediate positions.^[7]

In this paper I develop and defend a new intermediate position by maintaining that Spinoza is a neutral monist while rejecting core tenets of Wolfson’s subjectivism. My aim is to present an attribute-neutral reading that can accommodate Spinoza’s definition of God as “a substance consisting of an infinity of attributes” (E1D6). This definition raises a sharp problem for the attribute-neutral reading: how could a substance whose essence is attribute neutral be defined in terms of the attributes it possesses? I argue that this problem can be overcome by a precise analysis of the manner in which attributes depend on the intellect’s perception. Through a new interpretation of the role of the intellect in Spinoza’s definition of attribute (E1D4), my reading allows us to combine realism about the quantity of attributes with the denial of metaphysically distinct individual attributes, such as thought or extension.

I begin in section 2 by presenting the motivation for the attribute-neutral reading and raising the challenge of how to accommodate Spinoza’s definition of God as a substance consisting of an infinity of attributes. To make this challenge more concrete, I identify two interrelated questions that Spinoza’s definition of God poses for any attribute-neutral reading. The first concerns the plurality of attributes: if the essence of substance is neutral with respect to attributes, how can a substance have multiple attributes (e. g., the two attributes of thought and extension)? The second question concerns the infinite quantity of God’s attributes. If substance is attribute-neutral, how could a particular quantity of attributes (e. g., infinitely many) be the defining feature of any substance?

To address the plurality question, I propose a new reading of Spinoza’s definition of attributes (E1D4) in section 3, according to which attributes are representation-dependent. To motivate this reading, I draw on Spinoza’s own analogy between attributes and secondary qualities (like colors) in a letter to Simon De Vries (Epistle 9). I argue that in both cases, Spinoza believes that the content of our concepts has both an objective and a subjective component, as these concepts pick out an intrinsic property of their referent insofar as this property is perceived by a representer. This allows Spinoza to maintain that attribute concepts like ‘thought’ and ‘extension’ share the same objective component but differ in their subjective component, thus explaining how an attribute-neutral substance can have a plurality of attributes.

In section 4, I tackle the quantity question by arguing that Spinoza conceives of attributes under a particular variant of representation-dependence, which I call model-dependence. What is special about models is that they represent in virtue of their internal structure. Drawing on Descartes’ Geometrie, I argue that Spinoza’s attributes should be understood by analogy to model-dependent spatial dimensions, like breadth, depth, and length. The concept of ‘breadth’ derives its content from the role it plays in a representational model, meaning it has a subjective component, like secondary quality concepts. At the same time, the quantity of distinct dimension concepts enables the model to represent an intrinsic property of its object, namely, its dimensionality. Model-dependence thus allows us to explain how the attribute-neutral essence of God could be defined by a particular quantity of representation-dependent attributes. Although the essence of God is not really constituted by distinct attributes, its defining feature is that it can only be represented in its entirety by a representational model that distinguishes between infinitely many attribute concepts, just as a geometrical model must make an infinity of distinctions among dimension concepts to represent an infinite-dimensional figure.

2 The Attribute-Neutral Reading

Spinoza’s definition of an attribute as “what the intellect perceives of a substance, as constituting its essence” (E1D4) has been the central motivation for attribute-neutral readings of his metaphysics. Yet on its own, this definition does not entail that Spinoza has an attribute-neutral metaphysics: some scholars have argued that, for Spinoza, the intellect correctly perceives the substance as constituted by metaphysically distinct attributes.^[8] The appeal of the attribute-neutral reading is that it provides a straight-forward interpretation of the definition of attributes while also explaining the role that attributes play in Spinoza’s system. For example, Spinoza claims at E2P7S that “we find one and the same order, or one and the same connection of causes” across all attributes. This is typically referred to as his ‘parallelism’ doctrine.^[9] As Wolfson points out, this parallelism would be well-explained by the view that thought and extension “only appear to be two but in reality are one” (Wolfson 1958, 2:23).^[10] Moreover, Spinoza thinks that whether a mode is an idea or a body depends on how it is conceived, comprehended, or explained (E2P7), and the same applies to its causes and effects (E3P2). On an attribute-neutral reading, the conception-dependence of modes (and of their causal relations) is a natural outcome of the fact that thought and extension, which define the essences of these modes, are themselves dependent on the intellect’s perception.

However, many of Spinoza’s views appear incompatible with the neutral monist reading. For example, he repeatedly uses ‘God’ and ‘God’s attributes’ interchangeably (E1P19; 1P20C2), suggesting that God really is constituted by attributes. Moreover, Spinoza claims that all of the intellect’s ideas are adequate: “what is contained objectively in the intellect must necessarily be in nature” (E1P30D). This view seems to entail that if the intellect perceives the substance as constituted by attributes, the latter must really be so constituted (see Gueroult 1968, 1:XV). Finally, Spinoza is clear that God’s infinite intellect also perceives the substance – that is, God himself – as constituted by attributes (E2P4D).

If the attribute-neutral reading is to be credible, it must be sophisticated enough to accommodate these claims. As we have seen, however, Wolfson, the most influential proponent of attribute neutrality, defends a view that is incompatible with these texts. He maintains that the attributes are “modes of thinking” (Wolfson 1958, 146) constructed by the intellect and that such a mode “cannot have the essence of God as the object of its knowledge” (Wolfson 1958, 1:404). I agree with the near-consensus opinion that his reading is untenable. In addition to the texts cited above, it fails to explain why Spinoza would spend books II-V of the Ethics using attribute concepts when accounting for the metaphysics underlying human life, knowledge, and ethics.

Many interpreters have taken the shortcomings of the Wolfsonian paradigm to entail that attribute-neutral readings of Spinoza are untenable.^[11] But most of their objections do not apply to other attribute-neutral readings. For example, the attribute-neutral reading recently defended by Martin Lin (2019, ch. 4) says that different attribute concepts co-refer, picking out the very same essence of substance, but under different non-semantic guises.^[12] On this view, the difference between two attributes does not derive from differences in the substance itself, but only from the way the intellect grasps the substance (in the same way as the English word ‘city’ differs from the Greek word ‘polis’ not due to their meaning but due to their non-semantic properties). This allows Lin to maintain that the essence of substance is perfectly simple, as well as resolving long-standing puzzles about Spinoza’s monism, such as failures of substitution of co-referring terms in certain contexts. This attribute-neutral interpretation is not subject to most of the classic objections against Wolfson, as Lin argues in his replies to objections (Lin 2019, 91–92).

Nevertheless, one objection that has been repeatedly raised against the attribute-neutral reading has never been sufficiently addressed. This objection stems from Spinoza’s definition of God at E1D6.^[13] Spinoza’s entire metaphysical system depends on the view that “no substance can be or be conceived” other than God (E1P14). His proof that God is the only possible and conceivable substance is based on the following definition:

By God I understand a being absolutely infinite, i. e., a substance consisting of an infinity of attributes, of which each one expresses an eternal and infinite essence.^[14] (E1D6)

It is hard to overstate the importance of this definition in Spinoza’s metaphysics. Possessing an infinite quantity of attributes is the single defining feature of the only being that exists by the necessity of its own nature and on which all other beings depend.^[15] As I mentioned in the introduction, this definition gives rise to a pair of interrelated questions: the plurality question – how could an attribute-neutral substance have multiple attributes in the first place? – and the quantity question – how could a particular quantity of attributes be the defining feature of any attribute-neutral substance? Scholars have addressed the former in a variety of ways that preserve attribute neutrality, typically by claiming that the dependence of attributes on the intellect renders the plurality of attributes metaphysically benign.^[16] But this way of answering the plurality question makes the quantity question even more difficult and pressing, since Spinoza’s definition of God places the quantity of attributes at the center of his metaphysical system.

We can return to Lin’s neutral monist reading to see why the quantity question is so difficult to address. In order to account for the “simplicity” of substance with respect to attributes, Lin claims that distinctions among attributes “imply no metaphysical difference” in the essence of substance (Lin 2019, 101). This claim is appealing for a neutral monist reading that takes seriously the role of the intellect in E1D4. But as a result, Lin’s view seems unable to explain Spinoza’s appeal to the infinity of God’s attributes as a key premise in support of two of his central metaphysical conclusions: that God necessarily exists (E1P11) and that God is the only possible substance (E1P14D). The infinite quantity of God’s attributes could not support such clearly metaphysical conclusions if distinctions among attributes had no metaphysical implications.^[17] For this reason, Lin’s claim that attribute distinctions imply no metaphysical difference cannot be correct.^[18]

Lin’s interpretation does an adequate job of addressing the plurality question: co-reference under different non-semantic guises explains why distinctions among attributes do not threaten the metaphysical neutrality of the substance these attributes belong to. Yet the very same considerations that allow him to address the plurality question make it difficult to see how he could ever answer the quantity question. The challenge I take up in the rest of this paper is to develop a reading of Spinoza’s attributes that can answer both the plurality and quantity questions while maintaining an attribute-neutral account of the essence of substance.

3 Representation-Dependence and the Secondary Quality Analogy

Neutral monists and their opponents have disagreed on how to interpret Spinoza’s definition of attributes at E1D4, which says an attribute is “what the intellect perceives of a substance, as constituting its essence.” But to my mind, neither side has carefully unpacked the implications of the way the definition is actually formulated, and what this signifies in the context of Spinoza’s broader views about metaphysics, perception, and the intellect. The definition says neither that attributes are modes of the intellect, nor that they are intrinsic to the substance. Rather, the definition says that our notion of ‘attribute’ refers to the nature of its object insofar as the latter is being represented by an intellect. To unravel the nuances of Spinoza’s conception of attributes, we must therefore understand his views on the metaphysics of representation-dependence more broadly.

Philosophical discussions of representation-dependence have typically centered on secondary qualities such as colors. Whether an object is red depends on the perceptual representations it produces in certain perceivers; these representations partly explain why the object counts as red. The problem of the ontological status of secondary qualities, and their relation to the intrinsic geometrical properties of things, was one of the central concerns of seventeenth-century metaphysics. Thus, Galileo, in The Assayer, claims that qualities like color and flavor are “mere names so far as pertains to the subject wherein they reside, and that they have their habitation only in the sensorium” (Galilei 1957, 309). This is the same text in which he famously claims that the “grand book” of the universe “is written in the language of mathematics, and its characters are triangles, circles, and other geometrical figures” (Galilei 1957, 184). These two views are closely related: the fact that colors and other perceptual qualities are not reducible to geometry is one of the reasons they needed to be eliminated from mechanistic theories of nature.^[19]

Scholars have not discussed secondary qualities in the context of Spinoza’s definition of attributes, but he enthusiastically endorses their resemblance to attributes in Epistle 9.^[20] In this letter to Simon De Vries, Spinoza explains the relation of attributes to substance, and in particular his view that an attribute is so called “in relation to the intellect, which attributes such and such a definite nature to substance” (Epistle 9). He tells De Vries that the difference between the concepts ‘attribute’ and ‘substance’ is like the difference between two words that refer to the same object, and gives the following example:

By flat [surface] I mean what reflects all rays of light without any change; I understand the same by white, except that it is called white in relation to a man looking at the flat [surface]. (Epistle 9)

Confirming what is suggested by his definition of attributes in the Ethics, Spinoza explicitly states that attributes are analogous to representation-dependent secondary qualities. In section 3.1, I argue that Epistle 9 contains important insights into Spinoza’s metaphysics of representation-dependence. In section 3.2, I apply these insights to the plurality question I raised in section 2, arguing that representation-dependence enables Spinoza to maintain that substance is metaphysically attribute-neutral despite possessing a plurality of attributes.

4 Model-Dependence and the Dimension Analogy

The secondary quality analogy alone cannot show us how to answer the quantity question. This is because attributes are unlike secondary qualities in several key respects. Identifying these differences can allow us delineate criteria for a more accurate analogy. In section 4.1, I raise four points of divergence between Spinoza’s attributes and secondary qualities: arbitrariness, intellectualism, self-reference, and inescapability. In section 4.2, I argue that a different variant of representation-dependence, which I call ‘model-dependence,’ can bridge these differences. This opens the door for a more sophisticated analogy between attributes and model-dependent dimensional concepts like breadth, depth, and length. I argue in section 4.3 that this analogy can show us how to answer the quantity question while maintaining an attribute-neutral reading of Spinoza’s metaphysics.

4.3 The Dimension Analogy and the Quantity Question

As we have seen in section 4.2, the ‘geometric calculus’ proposed by Descartes begins with an arbitrary assignment of dimensional axes and unit lengths to a representational model. Once these choices are made, certain facts about the values of every point in the space follow. For example, every point along the x-axis has a y-value of zero, every point along the y-axis has an x-value of zero, and any figure on this plane can be precisely modelled in terms of the x- and y-components of each of its points.

The analogy between dimensional concepts, like ‘breadth’ and ‘length,’ and attribute concepts like ‘thought’ and ‘extension’ helps us see how an attribute-neutral reading can make sense of Spinoza’s claim that God’s defining feature is the infinite quantity of attributes he possesses. Just like attribute concepts, the distinction between ‘breadth’ and ‘length’ is entirely representation-dependent: nothing about a figure determines which region constitutes its breadth and which its length. Yet it is by distinguishing between two determinate spatial dimensions that a model is capable of representing an intrinsic property of this figure: its dimensionality. To illustrate the benefits of the analogy, we can consider two models that differ only in their arbitrary assignment of dimensional axes. They can represent the same figures, but the x- and y-values of the points that compose these figures will differ according to each model.

The terms ‘x-axis’ and ‘y-axis’ are analogous to the concepts ‘thought’ and ‘extension’ (or any other pair of counterpart attribute concepts).^[33] But we should notice that ‘x-axis’ is an ambiguous term: it refers to a different region of the space depending on which model we are looking at. To avoid confusion, we can index these terms to a model; ‘x-axis₁’ refers to the region labeled ‘x-axis’ in Model 1, and vice-versa. We can also use dimension-neutral terms to keep track of what is invariant across models: ‘x-axis₁’ and ‘y-axis₂’ co-refer to a region we can call R1, while ‘x-axis₂’ and ‘y-axis₁’ co-refer to region R2. Whether we refer to region R1 as an x-axis or a y-axis does not affect how we describe its intrinsic character. These terms differ only in their relation to the representational models (as indicated by the different subscripts). This representation-dependence is a feature dimensional concepts share with the secondary qualities ‘white’ and ‘nectargreen’ we considered in section 3. The dimension analogy therefore retains the answer to the plurality question we arrived at in section 3.2: like the dimensional concepts ‘breadth’ and ‘length,’ ‘thought’ and ‘extension’ do not pick out two metaphysically distinct properties, so the neutrality of the substance is not threatened.

Figure 1:

Two models of a plane.

Nevertheless, there is a crucial difference between these dimensional concepts and secondary quality concepts. While it is arbitrary which term we use to label the region R1, the fact that we use two of these terms does depend on an intrinsic property of R1: that it is a region of a space with at least two dimensions. By referring to a certain quantity of distinct dimensions, we are ascribing a certain dimensionality to the represented space, even though the dimensional concepts differ only in how they describe the representational model they belong to. At the same time, two-dimensionality is not representation-dependent: unlike ‘breadth’ or ‘length,’ the dimensionality of a space does not in any way depend on choices made in any representational model. It is an intrinsic feature of the space, yet it determines the quantity of representation-dependent distinctions that must be drawn in a model that represents it.

This analogy shows us that it is possible to render neutral monism compatible with Spinoza’s definition of God as a being consisting of infinitely many attributes. The difficulty in answering the quantity question stemmed from the worry that if all attributes are representation-dependent, then their quantity cannot imply anything about the intrinsic nature of the substance. The dimension analogy helps us to see that this is not the case with model-dependent concepts: the quantity of dimensional concepts we use in a geometrical representation corresponds to the dimensionality of the object we represent, even though the content of each of these concepts is representation-dependent.

Geometrical models deploy multiple representation-dependent dimensional concepts that differ only in their subjective component, yet the quantity of these concepts is determined by an intrinsic property of the represented object. This is how the dimensional analogy can help us answer the quantity question in Spinoza’s theory of attributes, where the secondary quality analogy could not. In the case of secondary qualities, the fact that ‘white’ and ‘nectargreen’ can both refer to the same flower carries no metaphysical implications about the flower’s nature (see section 3.3). In the case of model-dependent dimensions, the fact that 'breadth’ and ‘length’ can both be used to describe a region of a space implies that the space is (at least) two-dimensional. Distinctions among representation-dependent attributes can imply a metaphysical difference, so long as these attributes are model-dependent.

5 Spinoza’s Definitions

In section 4, I argued that the model-dependence reading of Spinoza’s conception of attributes has the resources to explain how an attribute-neutral substance could be characterized by the quantity of attributes it possesses. Given the clear benefits of the attribute-neutral reading (section 2), the ability to answer both the plurality and quantity questions constitutes evidence in favor of the model-dependence reading of attributes. In this section, I return to Spinoza’s definitions of attributes (E1D4) and God (E1D6) at the start of his Ethics, and outline the interpretation of these definitions that is implied by the model-dependence reading of attributes.

According to the reading I have proposed, when Spinoza defines attributes as “what the intellect perceives of a substance, as constituting its essence,” what he means is that individual attributes, like thought and extension, are defined by the role they play as components of intellectual models of substance. For something to be extended is for it to be represented by an intellectual model that includes the concept of extension as one of its model-dependent concepts, and likewise in the case of thought (or any other of the infinity of attributes). This is why Spinoza tells De Vries that attributes are so called “in relation to the intellect, which attributes such and such a definite nature to substance” (Ep. 9). This reading implies that individual attributes depend on how they are represented by a particular intellect and thus do not constitute the essence of substance as it is in itself.

Matters are different when we consider the quantity of attributes, as in E1D6, the definition of God. To say that God has infinitely many attributes is not yet to identify any one of them. Rather, this definition concerns how God can be intellectually represented, rather than how God is represented by some particular intellect. On the model-dependence reading, this definition implies that unless an intellectual representation deploys an infinity of distinct attribute concepts, it cannot be a complete representation of the essence of God.^[34] It also follows that any intellectual model, no matter which attribute concepts it deploys, can be a (partial) representation of the essence of God: as Spinoza says in his explication of this definition, “whatever expresses essence and involves no negation pertains to [God’s] essence” (E1D6Exp). These implications concern the intelligibility of God’s essence, without being indexed to any particular intellectual representation of God.^[35] As a result, this definition does not define the essence of God as representation-dependent, even though individual attributes like ‘thought’ and ‘extension’ are representation-dependent. According to the model-dependence reading, the infinite quantity of attributes is an intrinsic feature of God’s essence which does not depend on any representation of that essence. This opens the door for Spinoza to use the definition of God to draw metaphysical conclusions, as he does when proving that God necessarily exists (E1P11) and that God is the only possible substance (E1P14D).

6 Conclusion

The appeal of the neutral monist reading of Spinoza’s metaphysics derives from his definition of attributes and the role they play in his philosophical system. At the same time, there is no shortage of textual and philosophical tensions that arise from the claim that substance is metaphysically attribute-neutral. My aim in this paper has been to tackle one of the largest extant hurdles for the tenability of the attribute-neutral reading, by arguing that it can be rendered compatible with Spinoza’s definition of God as a substance consisting of an infinity of attributes. By interpreting attributes as model-dependent, much like spatial dimensions in Cartesian geometry, we can explain how an attribute-neutral substance could be defined by its infinite quantity of attributes. No extant attribute-neutral reading has provided a clear answer to this question, giving the model-dependence reading a clear advantage over the alternatives.

I have focused on the way the model-dependence reading can help us address the quantity question and thereby accommodate Spinoza’s definition of God under an attribute-neutral metaphysics. But given the architectonic role of attributes in the Ethics, this reading also has implications for other aspects of Spinoza’s system. For example, Spinoza’s claim that every being must be conceived under some attribute, on the model-dependence reading, implies that all intellectual representations take the form of representational models whose content derives from their internal structure. This insight into the activity of the intellect may help us better understand how intellectual representations are structured, under what conditions they are produced, and how their content differs depending on which attribute concepts they deploy. While I have not developed answers to these questions here, the constraints imposed by the model-dependence reading can help clear the way for a more nuanced attribute-neutral account of Spinoza’s metaphysics, epistemology, and theory of mind.