In 1908, when Henri Matisse exhibited in Paris thirty paintings, sculptures and drawings representing his work of the previous eight years, a disgruntled critic complained of the artist’s "unhealthy state of mind, overworked by search and ambition." If only he’d realized what was coming. Just three years earlier Matisse had burst into notoriety as the most radical of the Fauves, or "wild beasts," of painting; the 1908 exhibition showed that the 38-year-old artist was ready to take stock of his work without false modesty, the better to push forward into new terrain. His sense of accomplishment and his restlessness could hardly be disentangled. "I do not repudiate any of my paintings," Matisse declared in his "Notes of a Painter," published a few months later, "but there is not one of them that I would not redo differently, if I had it to redo."

Where most observers saw madness and aggression in Matisse’s work, his one great early defender among the critics, Guillaume Apollinaire, saw a "Cartesian master." "We are not here in the presence of an extravagant or extremist undertaking," Apollinaire argued. "Matisse’s art is eminently reasonable." The poet was being sly, knowing as he did that to be reasonable, or rather to put one’s reason into practice, can be a most extreme undertaking. And Matisse was just as crafty when he claimed, in implicit contradiction to Apollinaire, to disdain intellect as a guide to painting: "I believe only in Instinct." True, Matisse is the most intuitive of painters, yet to express his instinct cost him immense intellectual as well as physical labor. "Often behind one of these works," he explained to a Catalan journalist, "a dozen more have been undergoing evolution, or, if you wish, involution, from objective vision to the sensationalist idea that engendered it." Matisse’s insight, if it has been precisely transcribed, is extraordinary. It would have been more conventional to assume that art begins with sensation and is gradually elaborated to construct an "objective vision." But Matissean vision is just the opposite, moving from objectivity to sensation, or rather (and here comes a curious oxymoron) to a "sensationalist idea," apparently a sort of intellectualized sensation. It’s as if, in Blakean terms, one were to start from experience to achieve innocence.

Matisse’s paintings can appear to have taken form effortlessly. Their timelessness is akin to that of the icons that stunned the painter when he visited Russia in 1911—"the true source of all creative search," he declared. Of course, the frank evidence of multiple revisions observable on the surfaces of most of them tells us that this sense of ease is deceptive. Readers of Hilary Spurling’s biography of the painter can easily come away from it thinking of the artist as an absolute kvetch: high-strung and anxious, a reckless workaholic on the verge of a nervous breakdown. Of the 1906 painting The Gypsy, he later observed that it "shows the energy of a drowning man whose pathetic cries for help are uttered in a fine voice," while the mural Music (1910) he described as "an immense effort which has exhausted me." Of the glorious 1912 still life Basket of Oranges, later purchased by Picasso and now on view in Chicago, he told Françoise Gilot, "It was born of misery." When a painting happened to achieve an unexpected success he took little pleasure in it, seeing the work as just "the beginning of a very painful effort." A confirmed atheist—except "when I work"—he would use the Lord’s Prayer as a mantra to calm himself down. And yet as his friend the writer and socialist politician Marcel Sembat observed, "He has no wish to offer other people anything other than calm."

The war years made Matisse’s anxieties even more acute. He volunteered for military service and even bought combat boots but failed the medical exam; he appealed his refusal and was turned away again. His mother was caught in the German-occupied northeast; painter friends were in the trenches. Contributing prints to fundraising efforts for civilian prisoners of war did something to assuage his feelings of guilt, "sickened by all the upheaval to which I am not contributing." His living had been dependent on a market composed mostly of foreign buyers, now dispersed; there were no more commissions from Russia. But in his art, Matisse never thought of seeking safer ground; away from the front, his struggles became all the more acute. Alfred Barr, in his pioneering 1951 monograph on the artist—still the first stop for anyone wanting to look further into his work—defined the years 1913–17 as Matisse’s period of "austerity and architectonics." The perceptual puzzles of the Cubists having stolen the Fauves’ thunder as the last word in artistic radicalism, and with the fate of France in doubt, the luxe, calme et volupté, the bonheur de vivre that had given names to some of Matisse’s most ambitious canvases of the previous decade yielded, at least for a while, to a new sobriety of color and emphasis on geometric structure. "Matisse: Radical Invention, 1913–1917," at the Art Institute of Chicago through June 20 (and then at the Museum of Modern Art in New York City, July 18–October 11), puts the artist’s work of this period under a microscope—sometimes literally so.

Some works of this period verge on abstraction and, significantly, some of the most important among them were not exhibited until much later, as if the artist feared that the public would not comprehend them. When the 1914 View of Notre Dame—with its rough and hesitant approaches to an architectural volume that might have been unidentifiable without the title—was finally shown thirty-five years later, it was still seen by some as "an unfinished sketch to which Matisse had unaccountably signed his name." Another famous near-abstraction from the same year, French Window at Collioure, really might be unfinished—or perhaps it would be better to say that Matisse may never have stopped wondering whether he was done with it; in any case it was exhibited only after his death. The black rectangle that occupies most of the canvas is not simply a negation—the "black future" that the poet Louis Aragon later saw there—but a latency rich with possibility. A finely balanced uncertainty toward itself makes French Window at Collioure the quintessential Matisse painting—if there is one.

During these years Matisse produced some of his greatest works, and some of the best-known. The exhibition’s climactic next-to-last room includes three of his most monumental and mysterious paintings, The Piano Lesson and The Moroccans, both works of 1916 in MoMA’s collection, and the masterpiece completed the following year, Bathers by a River, from the Art Institute. Not that Matisse’s relentless self-criticism always stood him in good stead; his attempt to get the better of Cubism in an oversized 1915 pastiche of a Dutch still life in the Louvre is a thing of fits and starts. But a visitor who’d walked into the exhibition (or picked up the catalog) without reading its subtitle would never imagine that it simply covers the years 1913–17. Really, the exhibition chronology starts in 1907 (and there are even a few works dated earlier than that); you go through several rooms (or 140 pages of text) before reaching 1913. Why the discrepancy between the actual scope of the show and the more modest claim of the title? The reason, I’ll guess, was the unavailability of many of Matisse’s most important works from the half-dozen years before the war. Because his greatest collector in the prewar period was the Russian textile manufacturer Sergei Shchukin—could he have been the tired businessman Matisse so notoriously invoked as a potential receiver of his art in "Notes of a Painter"?—paintings like Harmony in Red (1908–09), Nymph and Satyr (1908–09), The Conversation (1909), The Dance (1910), Music (1910), The Painter’s Family (1911), Moorish Café (1912–13), Mme Matisse (1913) and many more are today in the State Hermitage Museum in St. Petersburg. None are in this exhibition. There are wonderful paintings from these years in Western museums too, not to mention some of the artist’s best sculptures. But while they are sufficient to illustrate Matisse’s development away from Fauvism, they can’t convey the whole story, so the curators were wise to refrain from appearing to claim any full accounting of the period.

But when Stephanie D’Alessandro of the Art Institute writes of Matisse having "radically changed direction" around 1913 (and in this she reflects the prevailing wisdom), I’m not convinced. In what sense could his paintings of 1913–17 be considered significantly more "austere" or "architectonic" than Bathers With a Turtle (1907–08)? Here the three nudes, starkly rendered, pensive, are isolated against an environment reduced to three imposing horizontal blue bands, an abstraction closer to the minimalism of Brice Marden’s Grove Group of the mid-1970s, or to the geometricized backgrounds of some of Matisse’s later works such as the still life Gourds (1916), than to the luxuriant foliage echoing the figure’s sensuousness in a somewhat earlier painting like Blue Nude (Memory of Biskra), from 1907. And the severity of Bathers With a Turtle is hardly a rarity in Matisse’s work before 1913. What about the 1911 Portrait of Olga Merson, in which the primly seated figure is hemmed in by a pair of blunt, almost brutal concentric black arcs—a painting in which geometry seems to be used less to "construct" a figure than to deface it with what Barr called a "scimitar-like stroke"? Even the works Matisse produced in 1912, a year spent mostly in Morocco—works that explore, as D’Alessandro well puts it, "the power of North African light to dissolve form and to reconstruct space into vaporous, mutable, interwoven layers of ethereal color"—are not as sensuous or exuberant as she seems to imply. Reflecting the strictness of Muslim mores as well as the intensity of North African light, the figures are hieratically static, self-contained, distant. Although the paintings include passages of delicious detail (for instance the pointillist shorthand fantasy of dots and dashes at the center of the sash worn by Fatma, the Mulatto Woman, from 1913) they generally use a mode of representation based on the principle of the adequacy of the least possible indication.

None of the paintings I’ve mentioned can be called typical. In fact, what’s amazing about Matisse’s production throughout this period is how restlessly experimental he was. Seemingly a natural dialectician, he was provoked by his every pictorial choice to wonder what would happen if he did the opposite. His inclination was not to balance out alternatives but to pursue each to its contradictory extreme: architectonics and arabesque, severity and sensuality, psychology and decoration. But if any single painting seems best to support the idea of a new sobriety entering Matisse’s work in 1913, it would be Woman on a High Stool, whose subject is Germaine Raynal, the wife of an art critic associated with the Cubists—although I can never look at the painting without thinking, instead, of Virginia Woolf. The painting is mostly a gray that mutates into chromatic highlights here and there—flashes of pink on the woman’s face and left forearm, blue and green in her skirt—while the heavy black outlines of the various forms at times become entirely schematic, for instance in the leg of the rusty red table behind her.

This seeming suppression of color was something new for a painter whose calling card since 1904 had been the fearless use of color. Again, though, it can hardly be called typical, but that Matisse did recurrently experiment with gray and black after 1913 is inarguable. French Window at Collioure and another painting of 1914, Portrait of Yvonne Landsberg, may be the most extreme examples here, but paintings like Head, White and Rose (1914–15); Goldfish and Palette (1914–15); Apples (1916); The Rose Marble Table (1916); Portrait of Auguste Pellerin (II) (1917); and Shafts of Sunlight, the Woods of Trivaux (1917)—not to mention The Piano Lesson, The Moroccans and Bathers by a River—are all ones that Matisse could not have made earlier because he would not have used black or gray so emphatically except to mark a contour.

In 1917 Matisse began to visit the aged Pierre-Auguste Renoir; they were close until the Impressionist’s death in 1919. Years later Matisse passed on to his younger colleague André Masson the story of the first time he brought one of his canvases for the patriarch to see. "Renoir exclaimed, ‘You must understand very well that I could never like such a thing! All the same you’ve put there a pure black that doesn’t disturb the light of your painting; if we’d done that it would have created a hole!’" But Matisse felt vindicated. "I’d won my point all the same," he told Masson. "The Impressionists had banished black from their palette; I put it back—and prominently—and a painter as in love with color and light as Renoir had the honesty to confirm it: Black is not only a color but also a light." This conquest of a color that’s not a color and a light that’s not a light is the inner story of Matisse’s art in the war years. His sudden intense production in 1913 of monotypes in which fiercely elegant white lines pierce voluptuous black fields—the two elements perfectly matched so that the delicate line is as substantial as the massive field, the black as luminous as the white—must have convinced him that something similar had to be possible in painting.

It was Gertrude Stein who best captured Matisse’s involutedly self-critical way of working: "One was quite certain that for a long part of his being…he had been trying to be certain that he was wrong in doing what he was doing and then when he could not come to be certain that he had been wrong in doing what he had been doing, when he had completely convinced himself that he would not come to be certain that he had been wrong in doing what he had been doing he was really certain then that he was a great one." If reading that gives you a headache, you’ve got the idea, and those who think of Matisse only as the artist who claimed to offer "a soothing, calming influence on the mind, something like a good armchair that provides relaxation from fatigue," should be warned that Stein’s words give a better idea than the master’s own of what’s in store at the Art Institute. To see the show properly demands effort, and rewards it; the catalog can be even more taxing. In exhibition and catalog, the curators—D’Alessandro of the Art Institute and John Elderfield of MoMA—ask us to follow Matisse’s working methods more closely, by becoming aware of that "evolution, or, if you wish, involution" of which he spoke, within each painting (through his process of revision) and from one painting to the next.

But there’s a danger to the method. Two dangers, rather: either losing sight of the painting being scrutinized to the welter of modifications and reconsiderations that the painter thought he’d subsumed into it, or, on the contrary, mistaking the painter’s revisions for so many moves in a chess game whose outcome he’d decided in advance. It’s worth quoting the catalog at some length to get an idea of just how minutely the curators have inquired into Matisse’s process. Here is what D’Alessandro reports about a figure in Bathers by a River, a painting from 1909: an X-radiograph "shows how deeply Matisse scratched into the paint as he adjusted the figure’s form. Examination with infrared reflectography makes the artist’s adjustments in pencil and dark blue and black paint even more clear. Here we can see at least five variations of the right leg, as the bather initially had a wider stance; her left leg grew heavier and then slimmer as Matisse worked toward its final tapered form; and her hunched back was reshaped at least four times." Furthermore, she continues, "An image of the work viewed under ultraviolet light emphasizes the successive layers of paint that Matisse applied over the initial figure as he reduced the torso and limbs on one side and built up the other."

The curators’ investigation of the paintings includes close ocular examination; the use of technical means to make more visible aspects that are already visible to the naked eye but otherwise easy to overlook or difficult to see clearly enough to describe properly; and then what might be called the third degree, the use of technical means to see things that are entirely hidden from view in the finished painting. In what may be their most ingenious effort, they have probed a series of photographs of Matisse in his studio in 1913, with an early version of Bathers by a River in the background. As the curators explain it, "software was developed to reregister these images so that the details of the painting, which were documented at oblique angles or otherwise distorted, could be flattened and aligned into the same plane; these were then stitched into a never-before-seen composite of the canvas at this critical stage." One might say that they have tried to use the tools of the conservator for the purposes of aesthetic analysis. But to what extent is this valid? At times the desire to draw back the curtain and reveal the hidden workings of the painter’s art seems downright prurient. After all, the surfaces of Matisse’s paintings already tell so much about his methods—and precisely all that he wished to tell. In 1946 the artist allowed himself to be filmed at work, drawing; in the finished documentary, the footage was shown in slow motion. The film amazed the philosopher Maurice Merleau-Ponty. "That same brush which, seen with the naked eye, leaped from one act to another," he wrote, "to try ten possible movements, dance in front of the canvas, brush it lightly several times, and crash down finally like a lightning stroke upon the one line necessary." You can see this film in the exhibition, and it’s everything Merleau-Ponty says, but it pays to remember Matisse’s reaction: "My hand made a strange journey of its own. I never realized before that I did this. I suddenly felt as if I were shown naked—that everyone could see this—it made me deeply ashamed." All those X-rays and infrared and ultraviolet lights would likely have mortified him as well.

Matisse said that only while painting did he believe in God. Merleau-Ponty put it a little differently: "The painter, any painter, while he is painting, practices a magical theory of vision." And Walter Benjamin: "Magician is to surgeon as painter is to cinematographer." As art historians, D’Alessandro and Elderfield are committed to an analytic inquiry that tends to dissolve pictorial magic; they approach the paintings like film editors trying to see their development frame by frame, or like surgeons trying to see structure by cutting into the skin; and yet their fascination with the minutiae of Matisse’s process, like the obsession of an unrequited lover, also tends to affirm the paintings’ aura. It sounds like a contradiction, but if anyone can teach us the innocence that succeeds "objective vision," it’s Matisse. D’Alessandro and Elderfield’s findings will surely be mined by their colleagues for years to come. For the rest of us, they’ve succeeded in making some of Modernism’s most familiar works seem, if not quite as maddeningly intractable as they could be to their maker, then nevertheless unresolved and difficult again.